by Carl V Phillips
Continuing from Part 1, I will skip the Executive Summary (the claims there are mostly redundant with those in the main body) and the Legal Authority (which actually does include some scientific claims), and pick up with the section, Background for Deeming All Tobacco Products To Be Subject to the FD&C Act. If time permits, I will circle back to the scientific claims in the previous two sections after I provide the context from the main body.
To start, I am simply cataloging the claims and responding to them in the order they appear. Particularly important points might be highlighted in the text, but the ordering does not imply importance. Synthesis of the implications will follow. Page numbers are from this version of the regulations.
Electronic cigarettes (or e-cigarettes), for example, are widely available in retail outlets such as kiosks in shopping malls and on the Internet and their online popularity has surpassed that of snus and nicotine replacement therapies which have been on the market far longer than e-cigarettes (Refs. 6 and 7).
It takes a few readings of this to realize it is probably technically true but misleading. Snus (which, of course, is just the Swedish word for moist oral snuff) is used by far more Americans than e-cigarettes. But its online sales are a much smaller portion of the total, and that is what this actually says. It is not clear why this particular comparison would matter, and it only serves to confuse the reader. Additionally, reference 7 is about product popularity, but reference 6 is a study of nicotine concentrations, which is clearly not a legitimate source for the claim being made.
Additionally, young adults often mistakenly think non-cigarette tobacco products are safe alternatives to cigarettes (Ref. 8).
Reference 8 is Richter et al., 2006. Yes, 2006, and based on data from 2002. At best, this could tell us about perceptions from more than a decade ago, which are rather out of touch with this fast-moving era we live in, not the present tense assertion in the text. The study is of a small population, consisting of focus groups from two southern cities. The products reported were all variations on cigarettes, along with “herbal smokeless”, whatever that is. The methods and results are so badly reported in the paper that it is hard to make sense of them, but basically the participants were asked to offer their guess about which products were more or less hazardous compared to regular cigarettes. The paper is number-free, so the results are fuzzy, but what is reported showed subjects rated the “non-cigarette tobacco products” (which include small cigars and shisha, as well as several cigarette variations that FDA already regulates) as more harmful than regular cigarettes about as often as they rated them as less, flatly contrary to what is implied in the above claim. If pushed to make a guess about which product is a little worse or a little better (keep in mind there was nothing in this study that would imply they thought they were radically different), most people will do it; they might be right, even if the epidemiology cannot confirm that.
Indeed, the main results really showed that the participants were pessimistic about alternative products. They mostly rated smokeless as lower risk (clearly true), as well as hookah smoking (true based on everything we know, though the epidemiology is weak). They mostly rated small cigars and other cigarette variants as higher risk (which might or might not be true, but the magnitude of the difference is relatively small, so it is a guess either way). In other words, the claim that they are mistaken, on average, is not supported by the reference.
Research has shown that youth are also particularly vulnerable to the appeal of novel tobacco products (Refs. 9, 10, 11, and 12).
References 9, 10, and 11 are all about hookah smoking. There is something rather odd about a >400 year old practice being described as “novel”. Perhaps it is just jingoism. Since hookah smoking is more about a social outing than a product use per se, it is unlikely to appeal to older people with families and regular working hours (unless, of course, they are culturally part of the centuries-old tradition). However, this really tells us nothing about tobacco products themselves. Perhaps such observations could justify regulating shisha, but they tell us nothing about the impact of “novel tobacco products”.
Reference 12 is about RJR’s dissolvable smokeless tobacco products (which are already regulated by FDA) and its primary finding is that there was a fair bit of awareness, but very little interest in using them. More important, the only age breakdown was 18-39 versus 40 and older. While some of us might like to think of a 39-year-old as a youth, it is clearly an inaccurate claim. Moreover, a mere 4% of those in the younger group had even tried the products.
Because of their addictiveness and the marketing and sale of these products and their subsequent use by youths, some non-cigarette tobacco products can introduce youth into a lifetime of addicted tobacco product use and related harms, including premature death (Refs. 13, 14, 15, and 16).
I will set aside the casual use of “addictive” for when claims about it are actually made. For now, note that it is being asserted without any antecedent definition or support.
Reference 13 is one of those countless innuendo-based and analysis-free reportings of the content of a few cherry-picked tobacco industry documents. It contains some information about what a few people may have once hoped to achieve from marketing efforts, but no actual information that supports the claims made in that sentence. Reference 14 reports that smokeless tobacco was advertised in magazines, some of the readers of which were under 18. It supports the claim that children saw (past tense) such ads. That is only vague support for the most minor step in the logic in the sentence. The “these products” logically refers to “novel tobacco products” from the previous sentence, but the products advertised during this period were not novel. Just because a few who saw these ads later used the products (“subsequent”) does not mean there was causation, as implied.
Reference 15 is a little closer to the claims in the sentence, but only if the authors’ assertions are blindly accepted. They really only show that the moist snuff market evolved with product improvements (mostly offering a wider variety of sub-brands) in the early 2000s. They claim that the availability of nicotine from the products increased, but their numbers actually show mostly noise, with some brands and subcategories going up and others going down. Then they claim that use by children and young adults has increased, though the magnitude was trivial, and imply without any basis that this was caused by their other observations. They ignore other obvious explanations, like the fact that during this period, more people were learning how low-risk these products are (despite the efforts of people like those authors). In short, that paper provides no apparent support for the quoted sentence, other than via the unsupported assertions of the authors.
Reference 16, also by rabid prohibitionists, merely shows that many young people who use smokeless tobacco also smoke (which in no way suggests the smokeless use is causing smoking for the reasons presented here). So none of those references provide support for the main claims in that sentence, that the products are addictive (absent from all), marketing was causing youth uptake (asserted in one paper but never analyzed), that such use caused a lifetime of use (absent from all), or that there were any resulting harms or deaths (absent from all).
Further, many of the products proposed to be covered by this rule are offered in fruit and candy flavors, such as chocolate and grape flavors, making them especially attractive to children and young adults. For example, from 2010 to 2012, one cigar company introduced grape, white grape, and blueberry flavors to its line of little cigars and cigarillos (Ref. 17). In 2012, a manufacturer of nicotine solutions for e-cigarettes introduced Mint Mocha and Spiced Apple Cider flavors for their e-cigarette solutions (id.).
Mint Mocha and Spiced Apple Cider are the supposedly youth-oriented flavors they could come up with? Someone should tell that to the organizers of every median-age-50 church social I have ever attended.
More seriously, there is no evidence presented (or existing, to my knowledge) that suggests that flavors are “especially” attractive to children and young adults. The cited reference studied only adults, using dubious methods, and merely reported a relatively modest greater use among the youngest adults compared to not-quite-as-young adults. Since they seem to be relying on it heavily, it is worth unpacking after quoting more that is attributed to it below.
The paper is a classic example of an editorializing indictment build around a few statistics. The introduction includes an absurd photograph that shows that some tobacco products and some candy products (gasp!) use similar color palettes and packaging (as if there are any colors and practical package sizes and shapes that are not used for some candy product). It is not clear whether FDA was influenced by the demonstrated personal politics of the authors, but it presumably influenced how the authors reported their results.
The first nationally representative study (derived from more than 4,000 young adults aged 18 to 34) to examine the prevalence of the use of flavored tobacco products following the 2009 FDA flavor ban in cigarettes found that 20 percent of tobacco users in the study currently use a flavored tobacco product (Ref.17). The most common flavored products include flavored pipe tobacco, little cigars, and hookah tobacco (id.). Research has shown that flavored product use is higher among 18-to-24-year-olds than 25-to-34-year-olds, and that sugar preference is strongest among youth and young adults and declines with age (id.).
All of FDA’s claims about flavor preference basically come down to this one study, Villanti 2013. It is based on survey data from 2012 that is in a proprietary dataset. It basically just shows that the use of flavored products (which really means more flavored or “characteristically” flavored since all products obviously have flavor) is common among those who used products where flavors are traditionally part of the mix (shisha) or where the category evolved primarily to meet pent-up demand for flavored cigarettes (little cigars). It actually shows a surprising lack of flavor use for smokeless tobacco and e-cigarettes (the latter perhaps due to the calendar time of the survey, and would have changed as more vapers shifted away from C-store disposables).
Including all categories of subjects who tried any tobacco product in the last 30 days, 18.5% of them tried a flavored product at least once. (Note that FDA reported this as 20%. Rounding is reasonable, so long as it is consistent. We shall see if it is.) Notice that this means that a vast majority did not even try a flavored product. The authors report results from a multivariate model; they never explain what the model was, and such reliance on a black-box model is another serious problem here. The results from that show a modestly higher rate of flavor trying among 18-24 year-olds compared to those 25-34. This seems to be the entire basis for FDA’s claims.
Since the data reflects merely trying even once during the period, this difference is easily explained by younger people being more likely to shift among products (rather than having settled on a single product); if any of those they tried were flavored, they would be counted. Younger people are also more likely to be in social groups that go to a hookah bar, which would very likely result in consuming a flavored product. The conclusion that the younger group prefers flavors more than the older group is clearly not supported, much less the “especially attractive” claim.
Interestingly, though the innuendo seems to be trying to paint a picture of dumb impressionable kids, there is a comparably strong association in which more-educated people were more likely to have used (i.e., tried) flavors.
The last phrase in the quoted passage cites the same study, which is only about usage patterns, to make an assertion about sugar preference, which of course such a study could not assess. This seems to be based entirely on one unsupported throw-away clause in the conclusions that asserts the claim. It cites three references which the FDA could have used as the basis for the claim but did not. I am guessing they never looked at them. Those papers are about younger people being more likely to prefer sweeter sucrose solutions (i.e., soda), which has little apparent relevance to the subtle differences in sugaryness among tobacco products, which are obviously very different from soda. Moreover, there is no analysis at all in the paper about sweetness (characterizing flavors may or may not be more sweet than “unflavored” products).
Such findings indicate that flavored product use may influence tobacco-use patterns in young adulthood, a critical period when lifelong patterns of tobacco use are often established (Ref. 17 citing Ref. 18).
The reason for the through-citing is that Reference 18 is a 1981 Philip Morris USA paper or memo of some sort that presumably the FDA authors never looked at (I cannot find a copy of it). Whatever its content, it obviously has very little relevance to today’s world, which has radically different products and usage patterns. (I would guess that any reference to flavors in it was about flavored cigarettes.)
[Update: Thanks to a reader (see comments) I got a copy of Reference 18. Not only is the based on data from two generations ago or more, and indeed the only product mentioned is cigarettes. The document is not searchable and I did not have all day to search it (though I suspect I have already spent more time looking at it than FDA did), but as far as I can tell, the only flavor mentioned is menthol and there is no mention of it influencing behavior. If we interpret the reference as referring to only the last clause of the sentence (which, of course, is not how most readers will take it, so that in itself is misleading), the entire basis for that seems to be in the premise of the document, not its content. That is, it the document reports data about teen smoking and in the introduction says “Today’s teenager is tomorrow’s potential regular customer” and “it is during the teenage years that the initial brand choice is made”. So not only does the actual content of the report support any of the quoted sentence, but the only way in which it supports the latter half is in the form of introductory observations, and even those are far more equivocal than the claim that is attributed to them: The “potential” statement, and associated observations that the decline in teen smoking portends a decline in adult smoking, is almost a tautology, and are just asserted, not analyzed. Moreover, it is clearly not relevant in today’s world, where initiation takes place at a much higher age. The statement about brands has nothing to do with which tobacco product, if any, someone uses.]
The above consists of FDA’s entire scientific basis for going on to ask the following loaded questions:
Given this initial data regarding the increased prevalence of flavored tobacco products following the 2009 flavored cigarette ban, FDA seeks comments, data, and research regarding the following:
• Aside from this proposed rule, what additional actions, if any, should FDA take to address the sale of candy and/or fruit-flavored tobacco products to children and young adults? For example, what data should FDA request manufacturers submit in new tobacco product applications to establish that flavorants either do not raise different questions of public health, in the case of SE reports, or are appropriate for the protection of public health in the case of premarket tobacco product applications?
• What is the likelihood that individuals who engage in flavored tobacco product use will initiate cigarette use and/or become dual users with cigarettes?
• The prohibition against characterizing flavors established in the Tobacco Control Act applies to cigarettes only. Consequently, when this regulation is finalized and other tobacco products are deemed subject to FDA’s tobacco product authority, the statutory prohibition against characterizing flavors will not apply automatically to those products. However, once they are deemed, FDA may establish a product standard prohibiting flavors in those products. FDA requests information and data that would support establishing such a standard.
[Update: I found myself thinking more about my note about how everything has a flavor. It turns out that FDA never defines what they mean by flavored in this document. There is an established almost-definition of “characterizing flavor”. That phrase is used in some of the legal sections of the proposed regulations, and to note that these are banned in cigarettes (except for menthol). Some of the uses of “flavor” have an accompanying adjective (like “fruit”) that make them well-defined. But many uses of the word in the document are fundamentally ambiguous.]