Smoking initiation getting older, what does it mean?

by Carl V Phillips

This is a bit tangential at the start, but it is a good science lesson and you will see how it relates to THR, particularly the common claim that there is no value in THR because everyone is just going to stop using tobacco soon.

I was monitoring the tweets from today’s tobacco control industry’s (TCI) meeting to celebrate the 50th anniversary of the U.S. Surgeon General’s report on smoking (you can find the tweets at #SGR50 if you are interested — you will need to scroll down to today to find the comments on the meeting because there will be a lot going on at that tag over the next week).  There was a lot to despise about it.  Primarily it consisted of a bunch of people who were not even born yet (and a few who were children) trying to take credit for the historical anti-smoking measure that mattered far more than any other: telling people about how risky it is.  This seems to be an attempt to distract people from the fact that what they are doing now is basically useless — except when it is harmful, as when they try to prevent THR.

Apparently several of the speakers took this celebration of a report on smoking as an excuse to attack low-risk alternatives.  A few others, notably Matt Myers of CTFK and the president of the American Academy of Pediatrics (AAP), made the claim that a large majority of smokers start as kids.  The thing is, smoking initiation in the U.S. is now almost exactly half adults and half children, as shown here.

It is not clear whether they are simply illiterate, not understanding verb tense (a large majority of current smokers started as kids, but that appears to no longer be true of those who start).  More likely they are innumerate (that would fit their historical bad science, certainly) and do not understand the basic epidemiologic concept of age/period/cohort.  It is actually quite simple, though it confuses a lot of people who have never had it explained to them when they try to understand events that are happening over time.  A cohort is a group of people born within a particular period (for simplicity, just call it a year); period refers to calendar time; age is obvious.

It turns out that if you fix two of those, it determines the third.  For example, if you want to look at people born in (cohort) 1970 when they are (age) 44, that would mean you are looking at (period) 2014.  (Yes, I know — more precisely you are looking at part of 2014 and part of 2015 for each individual, but we tend to ignore that for convenience.)  Most important, for understanding these things, is that if you are looking at a particular age range (say 12- to 17-year-olds) over changing time (as is done when we chart how many children are starting smoking each year, as in the above link), then you are talking about a series of different cohorts.  That is, it is about different (though in this case, overlapping) groups of people.

To make this more concrete, most current smokers are members of cohorts that started smoking during periods where smoking initiation was at younger ages.  But the current cohorts are not doing this, and therefore in the future more smokers will have started at later ages.  If one does not understand the concept of cohort (as is apparently the case for those speaking today), it might be difficult to understand this.

This also lets us better understand the significance of the fact that smoking initiation is shifting to older ages.  As the tobacco controllers like to brag when they are touting their accomplishments (and like to hide when they are demanding more funding), smoking initiation among children is definitely trending down.  In particular, the TCI uses this observation to claim that THR is not useful because smoking is disappearing (never mind those who are already smoking — they can just die from it).  But initiation among the population as a whole is almost keeping pace with population growth.  With an understanding of cohorts (and being able to not conflate them with periods), it is relatively easy to hypothesize an explanation for this.

Hypothesis: Aggressive anti-tobacco (not just anti-smoking) propagandizing of children is causing them to avoid it until they reach an age where they throw off the manipulations of their younger self.  At that point, however, the portion of the population inclined to try or use cigarettes or other tobacco products is barely changed.  That is, pushing initiation to a later age does not substantially change the behavior of the cohort in the long run.

To better understand this, consider two other behaviors:  If children are effectively discouraged from engaging in the risky activity that is American football, they will never play it; basically no one takes up football as an adult. But if children are effectively discouraged from engaging in the risky activity that is sex, the percentage of the population that will eventually become sexually active is barely changed.  These represent the polar cases of whether stopping early initiation stops initiation entirely, or whether it basically does not reduce eventual initiation within the cohort at all.

The TCI wants us to believe (and themselves desperately want to believe) that tobacco use is more like football — that anyone who avoids starting it as a child will never start.  But the data is suggesting it is more like sex — where the inclination is not substantially changed by delay.  That is, a relatively constant portion of each cohort that is interested (obviously, a much smaller portion for tobacco than for sex).

If the above hypothesis is correct, we would expect that causing a decrease in children initiating would cause an increase in young adults initiating a few years later.  This generally fits the data — not perfectly, but better than does tobacco control’s hypothesis that if you stop a would-be child initiation, that person will therefore never start.  (To maintain that fiction, they try to maintain the fiction that most initiation occurs in childhood, despite the very simple data that shows otherwise.)  There will be a reduction in the number of smokers in this scenario (due to the lower rate in the cohorts who are currently children), but not the cohort effect that tobacco control claims:  There will be about the same number of 24-year-old smokers each year in the future as there are now.

If the hypothesis is basically correct, there will presumably still not be perfect substitution.  The trend will be at least a little bit of the football effect, with some who avoid starting as children losing all interest before become adults. So the current tactics for blocking childhood initiation will reduce total smoking in the cohort, but it appears that it will not be very much.

Getting this right has obvious implications for THR.  First, it shows that while the 1964 Surgeon General report and related education efforts mattered a lot in lowering the inclination to smoke, the near-plateau that has been reached in initiation is not actually changing.  This is contrary to what one might naively conclude from the data on current children that ignores the rest of the cohort’s lifecycle.  Thus, THR remains the only proven method for getting smoking rates to drop much below 20% of the population.  Second, it is further evidence against the TCI’s demonic possession theory of why people smoke.  It shows that people use tobacco because they like it.  Just as delaying sexual behavior initiation does not cause people to not like sex, delaying tobacco initiation does not substantially change the number of people who like tobacco.  Thus having low-risk tobacco products will make a lot of people happier, even apart from making them less likely to smoke.