In a recent paper, FDA researchers sought to identify characteristics of U.S. adults who might be interested in low-risk tobacco products. Research on this could be potentially useful for helping people make good choices. However, this study was not.
The authors make no attempt to hide their political bias. The paper begins by hyping the tobacco industry’s “history of marketing new products as less harmful,” despite the fact that this has been impossible for quite a while. As an example, the paper leads with the standard “light” cigarettes trope, carefully glossing over the facts that his was four decades ago and that public health enthusiastically embraced this (unfortunately ineffective) innovation.
The only other example they present is the marketing of e-cigarettes as lower risk. They fail to mention that (a) such a claim is obviously accurate and useful and (b) it was never actually legal in the U.S. in the first place.
The research was based on a survey that asked whether someone would be interested in using hypothetical low-risk or “less addictive” products. The use of a hypothetical product is reasonable, since it focuses attention on the stated characteristic, avoiding positive or negative beliefs about an actual product. However, as with political polling, this tends to create a positive bias compared to a real product with its real imperfections. [A example of this can be found in similar research I did a decade ago, which showed that smokers were far more interested in a “hypothetical” product that was really a description of available smokeless tobacco products than they were in the real-life products.]
The use of a mailed survey, mostly about perceptions of cancer risk, probably caused substantial selection bias. Only a third of subjects who received it responded. More important, was the phrasing of the key question.
The hypothetical question did not say that the alternative product is low risk, nor that independent experts have concluded it is low risk. Rather it said “a tobacco product made a claim that it was less harmful….” Setting aside how a product itself can assert a claim, this is loaded phrasing: It implies that the only source for the claim is the people who are trying to sell the product, and also that the product is merely a bit less harmful than cigarettes, rather than approximately harmless.
It is not terribly surprising that this weak information left the majority of smokers uninterested, despite the upward bias created by not identifying the particular product. It might seem a little odd that even a few percent of never smokers found this inviting, but that can be explained by the typical rate of subjects misreading questions or answering randomly in long written surveys.
The FDA is specifically interested in the impact of allowing claims like this to be printed on packages, which explains the phrasing of the question. But it is naive to try to assess the impact this way.
If the FDA had allowed weak (obviously accurate) claims like this to be printed on smokeless tobacco packages, as several manufacturers have requested, a few smokers might have noticed them and been persuaded. But news reports and advocates writing about the labels would have much greater impact. The net effect of such buzz might be limited, or even negative, if the discourse were dominated by disinformation denying the claim. Alternatively, the impact could be hugely positive. Either way, it would have little to do with the reaction of smokers who stumbled across the text on the package.
The majority of what appears in the report are cross-tabulations of how other personal characteristics were associated with interest in the hypothetical products. In an apolitical study, this would just be busy work used to justify writing a whole article despite having only one meaningful comparison. It is difficult to imagine how any of those detailed results from this poor survey could be informative.
This, however, is a political study. Its apparent purpose is to create the illusion of knowledge to justify the FDA’s continuing refusal to allow truthful labeling. Interestingly, the authors offer the positive observation, “it is encouraging that current smokers are more interested than never or former smokers….” However, they continue with, “this difference alone does not indicate that marketing MRTPs [i.e., allowing low-risk product manufacturers to say their products are lower risk] would benefit population health.”
This is true only because their research tells us almost nothing about the real world. If their numbers are taken at face value, and smokers switched at five times the rate that nonsmokers started using the product (in reality, an absurdly low estimate for that ratio), the population health effects would be overwhelmingly positive, even in a population with a very low smoking prevalence. This is, of course, because the real-world alternative products pose approximately zero risk.
But this is where the value to the FDA of the busywork starts to appear. The authors go on to highlight population subgroups who are particularly “vulnerable.” The absurd decision to refuse Swedish Match’s application to put such a comparative risk statement on their snus packages was based largely on rhetoric about an unidentified vulnerable subpopulation faring worse if that were allowed. It is always possible to slice-and-dice a population until some group is found that suffers a net negative effect from any regulatory change. The good news is these authors could not manage to support the claim that a particular subpopulation really was “vulnerable.”
As is typical for tobacco control papers, the authors falsely assert they have no conflicts of interest (there are few greater conflicts of interest than being employed by an agency whose policies depend on studies producing particular results). It turns out they did not manage to produce results that fully support their conflicting interest. Still, it seems likely that this will be spun into a reason to continue to deny accurate labeling.