Thanks for taking the time to read and comment, and sorry I didn’t see this earlier! You bring up a couple of really interesting points that I’ll try to respond to separately.
Gateway effects
There are a couple of reasons I’m unconvinced by the gateway hypothesis, i.e. that the observed correlation between non-combustible use and subsequent smoking implies that the growth of the former is causing an increase in the latter.
An alternative explanation is that lack of adjustment for confounders (‘common liabilities’ between vapers and smokers, like living with family that smokes, smoking peers, or a higher propensity for risk-taking) is what drives the majority of the observed correlation. The meta-analysis on vaping that you linked makes exactly this point: they reviewed 11 original studies on the association between vaping and subsequent smoking, and found that only two of them made comprehensive adjustments for key confounders, with one of these having other serious methodological errors, and the last one showing a small effect size. Thus their conclusion is that “while there is evidence of a relationship between vaping and subsequent smoking, there is much less conclusive evidence for a gateway effect.”
If the gateway effect were significant, I’d also expect this to eventually show up in population-level data. Of course, this may be masked by other factors that are causing drops in the smoking rate at the same time, but attempts to detect the first while controlling for the other in England and the US haven’t turned up much so far. I haven’t seen a lot of data on other countries.
Finally, and this is a more informal point, no clear and plausible causal mechanism comes to (my) mind for why someone who wouldn’t otherwise have become a regular smoker would become one after and due to use of a non-combustible product. If I enjoy snus, lozenges, or vaping, why would I switch to a more expensive product with a much worse health impact?
Bias
I agree that this is a field where an unusually high degree of skepticism of claims is warranted. Because of the amount of money and lives at stake, many of the people doing research have strong feelings and/or financial incentives related to the policy implications of their findings; or they enter the field specifically because of an opportunity for funding or a preexisting belief about the danger or opportunity presented by non-combustible products. As Andrew Gelman puts it, “Tobacco research is a mess, and it’s been a mess forever.”
That said, If we were to disregard the work of everyone with some sort of conflict of interest, I don’t think there’d be a lot left to read, so I generally try to take a trust-but-verify approach. For example, Carl Phillips has received tobacco industry funding, but the article I linked doesn’t present original data but a logical argument I can follow and identify holes in, and the factual claims can be checked against other sources. Becky Freeman is a strident advocate for increased regulation of both combustible and noncombustible tobacco and a protégé of Simon Chapman, who has repeatedly stretched the truth both with regard to the risks and benefits of snus and vaping. Tobacco Tactics, the organization whose articles you linked, receives funding from Michael Bloomberg, who has led global efforts to ban flavored vaping and made numerous exaggerated statements about vaping. I don’t think these biases should disqualify the evidence either of them present from being taken into account, as long as their factual claims are thoroughly checked.
Tobacco company strategy
I think PMI’s actions are pretty consistent with an established business responding to the emergence of a disruptive new product cannibalizing sales of its existing one. They initially worked to shut upstart competitors out of the new market through regulation in their support of the PMTA process. They’ve since acquired innovative competitors in the vaping space like Juul and are trying to purchase the largest snus producer. Non-combustible products accounted for 30% of their revenue in the first half of 2022, and this percentage has been growing for years. As this shift continues, it makes more and more sense for them to continue to lobby for broader accessibility.
In addition, I’d speculate that while companies like PMI don’t want to say so out loud, they recognize that the total addressable market for tobacco could grow substantially if health concerns are alleviated. Not that long ago, almost half of the adult US population used tobacco, and many of the people that stopped since then did so mainly because of those concerns, so the non-combustible market could be even bigger than the cigarette market ever was.
I agree that a gateway effect in adolescents is unclear due to confounders. I think this uncertainty should motivate caution though, rather than a conclusion that the gateway effect is non-existent. What are your thoughts on the Adermark et al. (2021) systematic review?
I wouldn’t expect a clear pattern to arise in population data due to the opposing factors driving rates down, which you mention. However, those time-series on the U.S. and U.K. have led me to decrease my credence, so thank you for pointing me to them!
I confess that on the mechanism point, we appear to have completely opposing intuitions. I think there are very plausible social patterns that could be at work. A 13 year old vaping because some of her friends vape would be easy; smoking because some of her friends smoke less so. At 15, if she had been vaping for 2 years she would have experience buying, owning, sharing, and using tobacco and/or nicotine products—I would think it there would be much less of a barrier to her starting to smoke occasionally and then consistently, compared to the case where she never vaped. But these are just contrasting intuitions we have—maybe we could look into qualitative studies to calibrate ourselves better.
On bias, thanks for illustrating that the picture was more complex than I originally thought. However, I think I want to push back strongly on the idea that those pushing regulation have bias similar to those pushing against. Tobacco companies have an extremely strong financial motive, bordering on existential. Bloomberg, Freedman, Chapman, and others don’t have that motive, so far as I can tell. Your points about them mainly point to the fact that they have been pro-regulation for some time and at some volume, which shouldn’t be surprising, and perhaps with some methodological flimsiness. That flimsiness would be concerning, but I think it’s a quite different concern to the tobacco industry’s bias.
That’s a fair point that you were convinced by Philips’ arguments rather than his credentials or history. But I think in a report arguing for substantive action, and concerning a major counter-argument, it would have been better to explain and defend those arguments rather than just linking to them and saying that you found them persuasive. However, that’s a point of form and other readers may think differently.
Your final points on tobacco company strategy are interesting and thought-provoking. I certainly feel less resilient on my original point, especially since I don’t know corporate strategy well. On reflection, I think I am perhaps quite cynical, though I’d argue that cynicism is justified. Phillip Morris’ ideal world would be one with much more vaping and smoking, and their history in causing and then hiding millions of deaths has made me view them with great distrust.
Thanks again for your substantive and thoughtful reply, I think it’s likely my beliefs on this will shift with time!
Thanks for taking the time to read and comment, and sorry I didn’t see this earlier! You bring up a couple of really interesting points that I’ll try to respond to separately.
Gateway effects
There are a couple of reasons I’m unconvinced by the gateway hypothesis, i.e. that the observed correlation between non-combustible use and subsequent smoking implies that the growth of the former is causing an increase in the latter.
An alternative explanation is that lack of adjustment for confounders (‘common liabilities’ between vapers and smokers, like living with family that smokes, smoking peers, or a higher propensity for risk-taking) is what drives the majority of the observed correlation. The meta-analysis on vaping that you linked makes exactly this point: they reviewed 11 original studies on the association between vaping and subsequent smoking, and found that only two of them made comprehensive adjustments for key confounders, with one of these having other serious methodological errors, and the last one showing a small effect size. Thus their conclusion is that “while there is evidence of a relationship between vaping and subsequent smoking, there is much less conclusive evidence for a gateway effect.”
If the gateway effect were significant, I’d also expect this to eventually show up in population-level data. Of course, this may be masked by other factors that are causing drops in the smoking rate at the same time, but attempts to detect the first while controlling for the other in England and the US haven’t turned up much so far. I haven’t seen a lot of data on other countries.
Finally, and this is a more informal point, no clear and plausible causal mechanism comes to (my) mind for why someone who wouldn’t otherwise have become a regular smoker would become one after and due to use of a non-combustible product. If I enjoy snus, lozenges, or vaping, why would I switch to a more expensive product with a much worse health impact?
Bias
I agree that this is a field where an unusually high degree of skepticism of claims is warranted. Because of the amount of money and lives at stake, many of the people doing research have strong feelings and/or financial incentives related to the policy implications of their findings; or they enter the field specifically because of an opportunity for funding or a preexisting belief about the danger or opportunity presented by non-combustible products. As Andrew Gelman puts it, “Tobacco research is a mess, and it’s been a mess forever.”
That said, If we were to disregard the work of everyone with some sort of conflict of interest, I don’t think there’d be a lot left to read, so I generally try to take a trust-but-verify approach. For example, Carl Phillips has received tobacco industry funding, but the article I linked doesn’t present original data but a logical argument I can follow and identify holes in, and the factual claims can be checked against other sources. Becky Freeman is a strident advocate for increased regulation of both combustible and noncombustible tobacco and a protégé of Simon Chapman, who has repeatedly stretched the truth both with regard to the risks and benefits of snus and vaping. Tobacco Tactics, the organization whose articles you linked, receives funding from Michael Bloomberg, who has led global efforts to ban flavored vaping and made numerous exaggerated statements about vaping. I don’t think these biases should disqualify the evidence either of them present from being taken into account, as long as their factual claims are thoroughly checked.
Tobacco company strategy
I think PMI’s actions are pretty consistent with an established business responding to the emergence of a disruptive new product cannibalizing sales of its existing one. They initially worked to shut upstart competitors out of the new market through regulation in their support of the PMTA process. They’ve since acquired innovative competitors in the vaping space like Juul and are trying to purchase the largest snus producer. Non-combustible products accounted for 30% of their revenue in the first half of 2022, and this percentage has been growing for years. As this shift continues, it makes more and more sense for them to continue to lobby for broader accessibility.
In addition, I’d speculate that while companies like PMI don’t want to say so out loud, they recognize that the total addressable market for tobacco could grow substantially if health concerns are alleviated. Not that long ago, almost half of the adult US population used tobacco, and many of the people that stopped since then did so mainly because of those concerns, so the non-combustible market could be even bigger than the cigarette market ever was.
edit: typo
Thanks for such a detailed and considered reply!
I agree that a gateway effect in adolescents is unclear due to confounders. I think this uncertainty should motivate caution though, rather than a conclusion that the gateway effect is non-existent. What are your thoughts on the Adermark et al. (2021) systematic review?
I wouldn’t expect a clear pattern to arise in population data due to the opposing factors driving rates down, which you mention. However, those time-series on the U.S. and U.K. have led me to decrease my credence, so thank you for pointing me to them!
I confess that on the mechanism point, we appear to have completely opposing intuitions. I think there are very plausible social patterns that could be at work. A 13 year old vaping because some of her friends vape would be easy; smoking because some of her friends smoke less so. At 15, if she had been vaping for 2 years she would have experience buying, owning, sharing, and using tobacco and/or nicotine products—I would think it there would be much less of a barrier to her starting to smoke occasionally and then consistently, compared to the case where she never vaped. But these are just contrasting intuitions we have—maybe we could look into qualitative studies to calibrate ourselves better.
On bias, thanks for illustrating that the picture was more complex than I originally thought. However, I think I want to push back strongly on the idea that those pushing regulation have bias similar to those pushing against. Tobacco companies have an extremely strong financial motive, bordering on existential. Bloomberg, Freedman, Chapman, and others don’t have that motive, so far as I can tell. Your points about them mainly point to the fact that they have been pro-regulation for some time and at some volume, which shouldn’t be surprising, and perhaps with some methodological flimsiness. That flimsiness would be concerning, but I think it’s a quite different concern to the tobacco industry’s bias.
That’s a fair point that you were convinced by Philips’ arguments rather than his credentials or history. But I think in a report arguing for substantive action, and concerning a major counter-argument, it would have been better to explain and defend those arguments rather than just linking to them and saying that you found them persuasive. However, that’s a point of form and other readers may think differently.
Your final points on tobacco company strategy are interesting and thought-provoking. I certainly feel less resilient on my original point, especially since I don’t know corporate strategy well. On reflection, I think I am perhaps quite cynical, though I’d argue that cynicism is justified. Phillip Morris’ ideal world would be one with much more vaping and smoking, and their history in causing and then hiding millions of deaths has made me view them with great distrust.
Thanks again for your substantive and thoughtful reply, I think it’s likely my beliefs on this will shift with time!