under review at Tobacco Command
(29 Feb 2012)
Demonstrating that children believe that deadly tobacco products are harmless candy, an experimental study
C.V. Phillips, P.S. Heavner-Phillips, I.Q. Burstyn, P.L. Bergen
Populi Health Institute, Wayne, Pennsylvania USA
UASPH, Edmonton, Alberta Canada
FDA = United States Food and Drug Administration
NRT = Nicotine Replacement Therapy
OR = Odds Ratio
OTC = Over the Counter
QUANGO = Quasi-non governmental organization
SES = Socioeconomic Status
TCACS = Tobacco Control Alarm-Call System
UASPH = Unaffiliated Alberta Scientific Public Health researcher
Figures and Tables
Background: There has been concern expressed about children mistaking smoke-free tobacco products for candy. There is evidence that anyone doing a study on the topic will have an inappropriately great influence on policy, and meta-regression suggests an inverse association between study quality and impact on public health discourse.
Methods: Children of aged 15 months to 15 years were presented with a tableau of smokeless tobacco products, along with benign alternatives like candy, breath mints, dietary supplements, and medicine. A second experiment presented the younger children with sippy cups that contained water, nicotine solution (e-cigarette refill liquid), and other liquids. A third experiment presented the older children with selected products in de-branded original packaging.
Results: The children could not distinguish between tobacco products and candy (p=0.86). The younger children were more likely to reach for a cup of nicotine solution than a cup of water (OR=2.2645, p<.0314). When products were presented in their original packaging, subjects successfully identified medicine, but half the tobacco products were still most often identified as candy.
Conclusions: Smokeless tobacco products should be stored with the same care recommended for other somewhat hazardous items found in the household, rather than displayed on plates or stored with candy. Children should not be presented with toxic substances in contexts that send the message they should eat them, and should not be told to think of them as candy.
Dissolvable tobacco products consist of finely ground or powdered tobacco in a confectionary base that are held in the mouth while they dissolve, thereby delivering nicotine and perhaps other psychoactive alkoloids in a similar manner as other low-risk oral smokeless tobacco products. These include products that have been available in the United States for about a decade (Stonewall and Ariva lozenges), as well as several new products that have entered the market or are expected to do so soon (new lozenge and toothpick-shaped products from the tobacco companies RJ Reynolds and Altria). None of those products have sold widely to date. However, trend analysis by TCACS indicates that the major-brand products will gain in popularity. Moreover, one popular form of dissolvable tobacco product that is known by the street name "NRT" or "nert", which are the same as the others except that the nicotine is extracted from the tobacco and then put in the confectionary base, are already widely consumed.
There has recently been an explosion of organized concern about the risk of accidental poisoning of children who mistake dissolvable smokeless tobacco products for candy. It is already known that that young children may be at risk of ingesting small objects that come within their grasp. However, since tobacco is completely unlike any other consumer product, previous research on other products cannot be considered informative. The nicotine lozenge variants of dissolvable tobacco products have already been implicated in a significant number of toxicant exposure events, which suggests that there is risk that one of the tobacco company products might someday poison someone also.
Several previous studies demonstrate that there is a view that the public health community should encourage worry about the poisoning risk:
An anti-tobacco QUANGO of the Virginia state government calling itself "Virginia Foundation for Healthy Youth" issued a press release  reporting that when teens were presented with dissolvable tobacco products without any context, they often thought they were mints or candies. While it might be considered a minor limitation of the study that the methods and actual results of the study were never reported or that the research was designed and conducted by high school students without any apparent supervision by scientists, the subsequent wide coverage in the popular press confirms that it was credible, important, high-quality research.
The Indiana Department of Health has mined social networking comments and otherwise gathered comments from teens about dissolvable smokeless tobacco products. Some commentators have expressed concern that they have never reported their methods and their results appear to be unrepresentative cherry-picked out-of-context quotes. However, these concerns can be ignored because the importance of this research is demonstrated by the production of high aesthetic quality slideshows that have generated repeated invitations to report to government agencies and public health gatherings [e.g., 2].
Additionally, there is even one study where a vague description of the methods and results was published, though unfortunately it has still never appeared in a scientific journal . The study, from a research group calling itself "Harvard School of Public Health", was widely cited in the popular press. (Note that while this organization does not seem to be related to the prestigious research institution called "Harvard University", the allegations in the literature that it is actually a lobbying front group for shadowy government agencies and industry have not been proven, and so cannot be considered a reason to doubt their research. Thus, we consider their work to be just as credible as the other studies cited herein.)
News reports, based on the highly touted press release from the "Harvard" study authors, indicated there was a dire risk of accidental poisoning from "tobacco candy". One possible limitation of this work is the almost complete lack of correlation between the study results and the claims touted to the press (p=.984). For example, popular media reports based on what was communicated to the press by the research team leader include the claim that "600 children a year" are poisoned by "tobacco candy" [e.g., 4], though this differs (p<0.001) from the actual study results by 600/yr. This lack of correlation should be interpreted as further evidence of the seriousness of the public health worries. Creating fear as a substitute for evidence is a standard invocation of the Precautionary Principle, which should be applied in situations where it appears there will never be enough evidence to support the necessary warnings.
Related research from several government and activist organization has looked at children's perceptions of flavored (though not necessarily dissolvable) tobacco products. For example, work conducted by the Utah Health Department used an innovative method in juvenile research, in which a researcher disguised as Santa Claus first tricks the children into thinking they are in a benign situation, and then warns them about the dire risk the researchers just exposed them to .
There is evidence that these results are being taken very seriously. For example, a spokesperson associated with the Indiana studies was invited to present for half an hour at recent FDA hearings on dissolvable tobacco products, while the experts on the public health benefits of introducing new low-risk tobacco products were restricted to three minutes each. The lead author of the "Harvard" study is frequently invited to participate in policy discussions, including in some forums that normally restrict participation to people who draw only those conclusions that are related to their research results.
A group of US Senators recently wrote a letter to the FDA urging removal of a "dangerous loophole" about dissolvable tobacco products . Once again, concern was so great that the fact that the "loophole" was irrelevant to the concern (it was a legalistic trick by FDA to delay acting on a statutory requirement that they recognize certain products as less harmful than smoking) and the "solution" did not address the "danger" (the matter had no relevance to whether young children could be given access to the products) posed no obstacle to decisive action by respected politicians.
Given that concern about this matter is so intense even in the absence of supporting evidence, it is clear that this is an important public health issue. It is established practice in modern public health that the assessment of a risk is most reliably determined by those who express the most concern. This is especially true in cases where, if action is not taken immediately, the concern might be proven unfounded. Thus, a public health activist Code Gamma Alert has already been issued, including the monthly television news stories and invoking of the mandatory opinion-alignment rules for professors at public health schools and employees of public health advocacy organizations.
[CARL, WE ARE NOT SUPPOSED TO TALK ABOUT THE ACTIVIST ALERT SYSTEM. YOU SHOULD PROBABLY TAKE THIS OUT BEFORE WE PUBLISH.]
While further evidence may be unimportant now that policy action is underway, we were curious about whether the results about children confusing dissolvable tobacco products with candy could generalize. In particular, we were interested in whether they could be replicated in a scientific paper authored by real researchers with scientific skills.
A convenience sample of 42 children, aged 15 months to 15 years was recruited. They were divided into the "younger" (n=25) and "older" (n=17) subsamples based on a cutpoint of age 7 years, as measured by a validated birthday frequency questionnaire.
In the first experiment, a tableau was created that included a collection of the candy and candy-like items listed in Table 1.
Table 1. Product list
Dissolvable Tobacco Products
Ariva (Star Scientific)*
Camel Orbs (RJ Reynolds)
Nicorette Lozenges (GlaxoSmithKline)*
Nicorette Mini-Lozenges (GlaxoSmithKline)*
Stonewall (Star Scientific)
Medicines and Supplements
Pepcid AC 10mg*
Tums Extra-Strength tablets
Zicam Cold Remedy "homeopathic"*
M&Ms dark chocolate mint
SweeTarts giant chewy
Tic Tac mint*
All listed products were used in the first experiment. Those noted with * were also used in the third experiment.
Brand names are given for identification purposes only and do not represent an endorsement of any listed product.
Each subject viewed tableaus of 10 items. The first that was presented included positive and negative controls. The positive control consisted of a marshmallow Peep candy, chosen because of its history as a model species in research [e.g., 7]. The negative control was a rubber ducky. The other items were chosen at random in each trial, thus eliminating any bias that might have been created by the grouping of particular products. An example of one of the the tableaus, as presented to the subjects, appears in Figure 1. The objects were always arranged in an attractive pattern (e.g., the circle in Figure 1); this served no scientific purpose, but it amused us.
|Figure 1. Example of one random tableau from first experiment|
Note: Pictured items are, clockwise from bottom center (numbers correspond to order of list in Table 1): 14, 12, 11, 3, positive control, 9, 4, 7, negative control, 6, 13.
The younger subjects, for whom the former methodology would not be age-appropriate, were grouped into pairs and presented with a tableau. The were asked "candy! who wants candy? pick out a piece of candy!" Results were recorded for which items were reached for or picked up. After grabbing ceased, another tableau was presented and this was repeated until a total of 20 grabs were recorded. When some pairs of the younger children demonstrated hesitation about reaching for any of the objects, the protocol was quietly rewritten to require that the researcher reach for and pretend to eat a random (non-ducky) item until the children complied.
A second study expanded the target of the study to include e-cigarettes. The failure to ban these new devices, so far, demonstrates that more research is needed about them and their components. The younger subgroup were presented with six sippy cups, that contained (pictured from left to right in Figure 2) e-cigarette refill liquid (a 6mg/ml nicotine solution), water, vodka, antifreeze, fruit juice (white grape), and e-cigarette refill liquid (a 20 mg/ml nicotine solution). Inclusion of milk and soy-based formula were considered, but concerns about the allergies mandated their removal. Coke and Sunny Delight were also eliminated from the experiment because of the recent discovery that sugary drinks are the world's most dangerous public health threat. The cup(s) that each child reached for or picked up were recorded. Picked-up cups were immediately removed from the children's hands and returned to the tableau. Children that started crying at that point were removed removed and thanked for their contribution.
|Figure 2. Sippy cup arrangement|
Note: Contents ordered as listed in text.
No incentives were offered to the study participants, no matter how much they begged (though the Junior Mints and vodka mysteriously disappeared). All of the recovered study materials were disposed of following appropriate biohazard protocols. The liquids were returned to their bottles; most of the solid objects were consumed by the research team; the drooled-on products from the young children tableaus were autoclaved and put in the municipal trash (there was no budget for analyzing biological samples); the NRT products, considered too unpalatable to consume but too expensive to throw away, were saved for future research. A count of the tobacco and medicine products revealed that approximately all of them had been recovered from the children (the measurement error was at a level that does not affect the study results).
Statistical analyses were conducted using SAS (vers. 9.1.3, SAS Institute, Cary NC), the preferred choice for researchers who want to show off that they are real scientists, but do not understand the statistics they are using well enough to just do the calculations in the spreadsheet where they recorded the data.
During the first experiment, five of the youngest children had to be dropped (from the experiment). Two refused to participate, wandering away and/or crying. Two others were apparently unfamiliar with candy, and thus did not produce reliable results. Another seemed to be claiming that every object in the tableau was a cat.
The remaining 37 subjects completed the first experiment, and all members of the younger and older subgroups completed their other experiment.
Subjects in the older subgroup all correctly identified the negative control (rubber ducky) as something other than candy/food, and 16 (94.1176%) correctly identified the positive control as either "candy", "a Peep", or "those things are so gross". The younger children were more likely to pick up the Peep than the rubber ducky at a rate that was borderline statistically significant after correcting for latent confounding by race, gender, and SES (p<0.1). Older children were more likely to volunteer an observation that the experiment seemed to be designed to intentionally trick them (p-value for trend = 0.03 after correcting for mutations in the data).
In the first experiment, the rate at which the tobacco (82%) and other non-candy products (78%) were identified as candy was the same as the rate for candy products (86%). The faster grabbing younger children demonstrated a preference for the larger items, but the trend did not achieve statistical significance. However, at a statistically significant rate, the older children identified the Vicodin (which had a distinctive pill shape, including scoring for breaking in half) as non-candy. The older children also identified the Stonewall (which has a color and shape that are unusual for candy) and Pepcid (smaller than most candies) as non-candy more often, but the difference was not statistically significant.
In the second experiment, the younger children were significantly more likely to reach for a nicotine solution cup than any of the other liquids. Compared to water, the odds ratio for reaching for the nicotine was 2.2645 (p<0.0314). The other liquids were chosen at rates that were statistically identical to the water.
In the third experiment, all of the older children recognized the positive control as a tobacco product and the Vicodin (in a prescription bottle) as medicine. For other non-candy products, most identified them as something other than candy: 94% for the Ariva and for the Pepcid (both are in a distinctly medicinal blister pack and do not look like typical candy). 65% identified the Zicam (in a distinctive supplement pill style OTC bottle) as non-candy and 35% as candy; given that it is a fruity sugary tablet, sold as homeopathic remedy, it is not clear who is correct, and in retrospect we realize that its inclusion in the study introduces unnecessary ambiguity.
Most important, for several tobacco products with the labels hidden many subjects still thought they might be candy, including 53% for Orbs (in a unique package, clearly meant to be childproof, but not similar to packages normally used for medicines) and 94% for the Nicorette products (non child-proof packages that are quite similar to mint or candy packages). One subject appeared to subconsciously mimic the Nicorette television ad where a consumer jauntily pops open the plastic tube with his thumb and smoothly pours a lozenge into his mouth one-handed. (Note: fearing this possibility, we replaced the content of the Nicorette packages with Tic Tacs and Mentos; since this part of the experiment was focused on the packages and those candies are almost visually identical to the tobacco products, we believe this precaution did not bias the results.)
The accurate identification of the positive and negative control items proved that our methodology was scientifically valid. Both older and younger children can differentiate candy from other objects, and older children can clearly identify tobacco products that are not shaped like candy. Given the reliability with which products can be made to not look like candy, even to young children, the persistent refusal of tobacco companies to employ non-candy shapes (e.g., plastic water fowl morphology) is evidence that they are marketing to children.
Our results replicate the previous research that showed that children can be convinced that non-candy tablets/lozenges are candy when they are presented in a context that suggests candy and lacks any clues that some products are not candy. This is the first study to demonstrate that children apparently cannot distinguish dangerous nicotine solutions from water or other liquids when they are presented in translucent vessels. It follows immediately from these findings that all smoke-free tobacco products that could lead to accidental poisoning should be banned, unless they have been approved by the FDA as a smoking cessation aid. In particular, those who use e-cigarette refill liquid -- "modders", as they are called in gang patois -- should not be allowed to continue to exploit a loophole that allows adults to legally possess poisonous liquids so long as they are not explicitly marketed as children's beverages.
Our findings suggest that package design might reduce the risk of mistaken identity and accidental toxification events. In other areas of consumer product regulation, there is a belief that in such circumstances, clear labeling and distinctive packaging should be required for potentially toxic products, with shapes and markings that make clear they are not food. For example, minimalist plain labels that would be similar to the obscured labels that we used to trick the children are generally forbidden. However, because tobacco is unlike any other consumer product, different standards should apply.
Nicotine lozenge manufacturers should not be required to change their packaging: If smokers were not able to flip the attractively-shaped bottle open with one hand and stylishly pour one (or all) of the lozenges into their mouths, then these products might no longer be considered "cool". That could, in turn, interfere with their proven ability to increase users' chance of quitting smoking by 0.0027%.
Packaging-based regulation is also an inappropriate approach for tobacco industry products, since reducing risk is never acceptable for health-affecting choices when prohibition is an option. Moreover, there is no evidence that younger children understand the cues from packaging. While it is possible that children might be less likely to try to drink nicotine solution if it is not in a sippy cup, there is no scientific proof of this claim. The only proven method of reducing the risk from these products is eliminating them entirely.
Strengths of this study compared to previous research on the subject include the fact that the authors understand irony and are not lying fanatics. Moreover, only one of them is functionally illiterate, and even he apparently understands scientific inference better than many of the authors of previous studies. Additionally, we have introduced to this field, for the first time, the methodologic innovation of not sending out a press release containing conclusions that are in no way supported by the content of the study. We consider this to be an important scientific breakthrough in itself, and suggest that future tobacco product researchers attempt to replicate the methodology.
This study has a few possible limitations. The older children proved surprisingly adept at detecting that they were being manipulated for some political purpose. This may have resulted from selection bias, since this work was carried out in a community of educated professionals with high-quality schools. The same demographic selection bias may explain why some young children were not practiced consumers of candy. If the experiment were repeated in a community where most of the teens were poorly educated, raised by public health officials, or otherwise never taught to think critically, we expect the results would be different. This would further strengthen our conclusions.
An additional factor that might be considered a limitation of the study, though it provides several important advantages in the context of tobacco research, is that we did not actually do it. That is, the methods narrative is pure fiction and we just made up the data (however -- and we realize it is difficult to believe, given that even one of the authors of this paper was sure we were making it up -- the literature review in the introduction is all true, down to the name of the author of reference #2). There were multiple reasons for not actually doing the study. First, we do not have access to a Romanian orphanage or a sample of foster children, and no halfway decent parent would ever allow their children to participate in such research. Second, we did not want to ruin perfectly good sippy cups and the Russian member of the research team would not allow us to waste vodka.
Third, we would never conduct experimental research on children without getting human subjects ethics approval, and no legitimate ethicist would ever consent to an experiment like this. Human subjects ethics requires a balancing of possible costs to the subjects against the potential benefit of the knowledge gained. Since it is obvious to any casual observer that it is possible to trick children into thinking that small pill-like items are candy, and to generally concoct situations that create confusion about objects that look, are scented, or are flavored like comestibles (be they medicine, cosmetics, tobacco products, air fresheners, industrial chemicals, etc.), there is no knowledge to be gained by these studies. While the risks of such a study would be reasonably low (assuming the children were carefully monitored and restrained from consuming any of the products), the mere act of including children in research, as well as taking actions that might convince a group of children that a collection of non-food objects are candy, creates a cost. Thus, such research could never be ethically justified.
Fortunately, such concerns did not create fatal limitations. Even though we did not recruit subjects or make observations, we carefully predicted what imaginary subjects would have said and done. While this may have created some confounding and measurement error bias in the results, it is well known in epidemiology that when a result is really strong, like most of ours are, it cannot be explained by study errors. Additionally, we conducted a sensitivity analysis that proved that our results are would have been similar had we actually gathered data (details omitted).
Moreover, in social science research related to tobacco, the drawing of conclusions based on how the researchers think subjects should act, rather than actual observations, is an established and accepted practice. For example, it is well documented that discussion and conclusions sections are frequently written before any data is collected. Furthermore, even patently false statements are so common in this field's research reports that it appears that they are mandatory. Thus, neither our methodology nor the fictional statements in this paper are outside the bounds of anti-tobacco research. Indeed, had we actually carried out the field study and adhered to legitimate scientific standards in our analysis and conclusions, our results would have been less useful, as such methodology would render our work incommensurate with the existing public health and tobacco research literature.
This study provides clear evidence that children can be persuaded to consume dissolvable tobacco products. The risk is clear even though -- in contrast with nearly every other toxic or small object in the world -- it does not appear that, so far, any children have suffered serious harm from these products (other than the NRT versions), or that they have frequently mistaken them for candy when encountering them in natural situations. Because they pose a risk, these products should be banned, as is appropriate for any product that causes risk but serves no necessary medical purpose.
Smokeless tobacco and e-cigarettes are not a safe alternative to smoking, and smokers should use only clinically proven cessation methods like NRT, social stigmatization, and lung cancer. Giving adults legal access to low-risk, affordable, appealing, satisfying nicotine products creates the risk that they might decide they are better off using nicotine rather than being abstinent, but without offering any profits to the pharmaceutical industry. Such a risk is not considered acceptable by public health authorities, even if the availability of the products might cause thousands of smokers to quit smoking. Tobacco harm reduction is a serious threat to public health officials and other career activists, and the tobacco industry must be stopped from making misleading claims such as "some nicotine products are much lower risk than cigarettes" and "citizens in our society have free choice".
Additionally, we conclude that it is unethical and dangerous, and should be criminal, for anyone to try to create confusion between dissolvable tobacco products and candy. This would include such actions as a parent putting the tobacco products in a candy jar, or activists constantly referring to the products as candy and encouraging the popular media to do the same.
Finally, we conclude that anyone who would actually carry out an experiment on children like the one described herein is in violation of basic rules of human research ethics, and may be guilty of criminal offenses. There is no evidence that previous researchers who carried out and published similar experiments sought human subjects ethics approval; indeed, it seems impossible that they would have gotten such approval. We recommend that national and local authorities consider whether sanctions are appropriate against the authors and sponsoring organizations.
Acknowledgments and author contributions: Support for this study was provided by Blue Cross (paid for the pharmaceuticals), Populi Health Institute (paid for the candy), and random donations of products from companies and advocates received over the years. CVP conceived of the study and wrote the manuscript. PSHP watched the process with fascination, contributing random strings of characters to the manuscript; he is responsible for any typos that appear in the text. IQB contributed statistics and words; he offered some amusing suggestions for embellishing the stories about the previous research, and was utterly floored to learn that we could not make those changes because the reported background was actually an accurate description of reality. PLB contributed content to the manuscript. The following bears repeating: None of the authors every seriously considered carrying out the human subjects research described herein; to do so would have been criminally unethical.
1. Virginia Foundation for Health Youth (2010). Press Release: More than 1/3 of Teens Surveyed ID Flavored Tobacco as Candy.
2. M Spitznagle (2012). Indiana’s Experience: Marketing of Dissolvable Tobacco Products. Invited slideshow, FDA Tobacco Products Scientific Activism Committee (TPSAC), available at http://1.usa.gov/yuB1MO
3. GN Connolly, P Richter, A Alequas, TF Pechacek, et al. (2010). Unintentional Child Poisonings Through Ingestion of Conventional and Novel Tobacco Products. Pediatrics 125:896-899.
4. The Week (2010). Fact Sheet: What is 'tobacco candy'? Available at theweek.com/article/index/202095/what-is-tobacco-candy
5. M Cortez (2010). Santa demonstrates danger of 'sweetened' tobacco products. Deseret News, available at http://www.deseretnews.com/article/705362953/Santa-demonstrates-danger-of-sweetened-tobacco-products.html
6. J Merkley, S Brown, T Harkin, B Mikulski, et al. (2011). Senators to FDA: It's Time to Close the Door on Tobacco Candy. Available at http://www.merkley.senate.gov/newsroom/press/release/?id=ea7b14e4-c4aa-490f-bc98-19713218acaf