As we note in the main text, addiction is not a well defined term. Most people feel that they know it when they see it, and their intuitive definition is probably as good as any. It turns out to be quite difficult to find a definition that discriminates between addiction and just enjoying something so much you want a lot of it. Other words like "compulsion" or "obsession" also have slightly different meanings, and many times addiction is confused with them.

Sometimes people seem to think of addiction as having to do with consuming a lot of something, even though it is bad for your health. But this definition is a little weak, because if you were able to make cocaine use entirely safe, you would still say someone who used a lot of it was addicted.

Some economists have suggested a definition of addiction that has to do with wanting more of something the more you consume, rather than becoming satisfied and bored by it, as we might with most things. Since economics includes the study of how people choose what to consume, it is not surprising that economics, rather than neuroscience, offers what is probably the most convincing definition of the term. This definition (the details are somewhat technical) distinguishes between addiction and compulsion, or merely liking something a lot. Under this definition of addiction and most others, it is possible to be addicted to exercise, work, listening to music, or any number of other things that we would not tell people to quit, just because they are addicted.

Because of the confusion surrounding this term, we prefer just to say that some people cannot or will not quit smoking (or using nicotine). It really only matters what people do, not what label we put on it.

Cardiovascular disease

This is health science jargon for diseases of the heart and circulatory system, including heart attacks and stroke.

Fifty times the risk?

Here is a technical analysis of the 50-fold risk claim ( for a shorter version see Brad Rodu's discussion here).

The claim that smokeless tobacco increases the risk of oral cancer by 50 times is such a remarkably persistent error that it calls for a more detailed analysis. The claim is particularly remarkable since there is an extensive scientific literature that shows that any claim that smokeless tobacco increases oral cancer risks by 50 percent (that is, by a multiple of 1.5) is extremely difficult to justify, let alone a multiple of 50.

This claims seems to have its origin in one figure reported in the Winn study that we discuss elsewhere. That number was part of an analysis intended to demonstrate what is called "biological plausibility". The idea is that if the smokeless tobacco product used by the study population was causing them to get cancer, we would expect the cancer to be concentrated at the locations where the tobacco was held (on their gums and inside their cheeks), as opposed to other parts of the mouth (tongue, roof of the mouth, etc.). We would also expect that those who had used more of it to have a higher risk of cancer (what is known as the "dose response").

So, this particular analysis separated out cancers of the gum and cheeks, and also divided the population into categories based on how long they had been using smokeless tobacco. The result was that for these particular cancer sites, the subjects in the category that had used the longest, when compared to non-users had almost 47.5 times the incidence of cancer (the number seems to always be rounded up to 50 when reported). But there are several reasons why this number clearly cannot be interpreted as showing that using smokeless tobacco increases your risk of oral cancer 50 times.

First, the clearest reason is that in the column just to the left of the one containing that number is the risk for other (not gum or cheek) oral cancers. That column shows that there is not an elevated risk for all other oral cancers. Remember that this analysis was designed to show that the excess cancer risk was just where we might expect it if it was caused by smokeless tobacco, and not elsewhere. What you probably do not know is that gum and cheek cancers are a fairly small fraction of all oral cancers. This means that even if the risk for these cancers were increased by a factor of 50, the risk for oral cancer as a whole would not be nearly that large because that increase would be the average of a large increase for some oral cancers and no increase for other oral cancers.

Second, the subjects in this category had been using smokeless tobacco an average of more than 20 hours per day for five or six decades. Again, a purpose of the analysis was to demonstrate that if there is a risk, it is greater for people for whom we would expect it to be greater, and this is certainly them. It is clearly inappropriate to conclude that the risk for people with far less total usage is anywhere close to this large. Indeed, for many exposures, a large exposure creates a substantial health impact, while small exposures seem to have no effect at all, so it is always difficult to extrapolate results like this. The total exposure that produced this figure is so much greater than the vast majority of smokeless tobacco users will have in their entire life that it is clearly not correct to conclude that more typical patterns of use cause similar risks (even for gum and cheek cancer).

Third, the results from this study in general appear to not be relevant to modern smokeless tobacco products. Subjects in this study used smokeless tobacco in a form very different from current use (they used dry snuff, produced in the early- and mid-20th century, rubbing it over most all of their gums, and leaving it there almost all the time). While studies of modern products show the health risks to be very low, there are some reasons to believe that older dry snuff products may have caused greater risk for disease. Thus, even to the extent that this study found an elevated risk, it does not appear to tell us much about current products.

Finally, a somewhat more technical criticism is that the statistical method used to generate this estimate appears somewhat suspect, and may have been designed to artificially increase the result. A group of subjects was excluded, supposedly because the quality of their data was not as good (though this is not so clearly true); if they had been included, the resulting estimates would have been much smaller. Similarly, the study included data on how much total smokeless tobacco someone had consumed in their life, but the authors chose to ignore this and report only the number of years of use; a similar analysis based on total consumption yields lower estimates. The division of the subjects into different use categories could have used different cutpoints, which also might have produced lower estimates.

(It is interesting to note that Winn's PhD dissertation which reported the results of her analysis methods in great detail, did not include the analysis that is discussed here (the published version of the study appeared after she graduated and started work for the U.S. National Cancer Institute, and was co-authored with her new employers). Since the published version of the study did not include nearly the details of the dissertation, it is not possible to figure out exactly what the authors did.)

A general caution about numbers like this: The repeating of this 50-fold number, in addition to being a misinterpretation of the evidence, would be misleading no matter what. To say something like "for some levels of use, the risk can be as high as 50" is not a legitimate way to interpret the science. For any exposure and disease, it is possible to pick out one subgroup of subjects from one study and get almost any number you want. Had Winn et al. analyzed their data slightly differently, everyone might be repeating the claim that there is a 40-fold increase or a 32-fold increase. From this it should be obvious that repeating any such number as if it were some kind of scientific constant is just wrong. Numbers like this are simply not meaningful outside of their very specific technical context.

To further emphasize the point, consider this: If you chose the right subpopulation from any study, including Winn's, you could report "for some levels of use, there is zero increase in risk" or even "for some levels of use, smokeless tobacco reduces your risk by 30%". Obviously no one is claiming the latter of these, but it follows from the same (faulty) logic that leads to the 50-fold claim. You can get either result by treating one particular statistical analysis of one subgroup of subjects as if it represented a universal truth.

Another simple scientific observation: If the risk of oral cancer from smokeless tobacco use were really 50-times, since a few percent of the population uses smokeless tobacco, we would expect that a huge portion of all oral cancer cases would be smokeless tobacco users. This is not the case; only a small fraction of people with oral cancer have ever used smokeless tobacco (just as we would expect by coincidence). Furthermore, there would be no controversy about whether smokeless tobacco causes any measurable increase at all, as there is. Every study that anyone ever did on the topic would show a huge elevated risk. It might vary some across studies, sometimes a 100-fold and sometimes 25-fold, but it would never be anywhere close to the actual results that studies have reported.

Inveterate smokers

Inveterate smokers simply means smokers who either cannot quit, or who don't want to quit.

Nicotine replacement therapy (NRT)

Nicotine replacement therapy (NRT) is a term commonly used to describe what we call pharmaceutical nicotine products.

Even though the term is popular, we avoid it because it is potentially confusing. The term comes from the use of nicotine patches, gum, and other products as a method for helping smokers quit using nicotine entirely. Because we are interested in exploring the use of long-term substitutes for smoking for those who cannot or will not quit using nicotine, we find the term to be distracting. The word "therapy" suggests physicians and medicine and other unpleasantness, and is not a good way to describe something people might want to use. It implies that nicotine users have a disease that needs to be treated. We prefer to be less judgmental, trying to focus on finding low risk alternatives that smokers might voluntarily switch to, rather than treating smokers as having a sickness that we experts and clinicians need to fix for them.

Moreover, the term is a bit odd, since "replacement therapy" usually refers to medical interventions that replace something the body is missing (like a hormone or water), not to alternative sources of something people choose to use. Finally, it is useful to keep in mind that the products are from pharmaceutical companies, with all that implies about marketing, profits, research funding, regulation, and other political and economic issues. Thus, we choose to use the more accurate and descriptive term, "pharmaceutical nicotine products" (which is defined below).

Pharmaceutical nicotine products

These include a variety of medical devices that provide nicotine, such as nicotine patches and gum, which are available without a prescription in many places, as well as a variety of inhalers, nasal sprays, topical gels, and lozenges which are generally available. These products, designed primarily to help wean smokers off of nicotine entirely, are usually called nicotine replacement therapy, though we find that term misleading. These products are sometimes discussed as a possible reduced-harm substitute for smoking.

Quit or die

Harm reduction advocates (like ourselves) often characterize the anti-tobacco-harm-reduction position this way. The usual messages to smokers are that smoking kills, and the only way to avoid this risk is to quit using nicotine entirely - there are no other choices, like switching products, cutting back, or anything else. What the message boils down to is "quit, or die". While this simplification seems a bit harsh, it is remarkable how often anti-tobacco and anti-harm-reduction advocates come close to saying exactly this.


Snus is the Swedish word for moist snuff, a product that is much more popular there than anywhere else in the world. Because the term "smokeless tobacco" is awkward and the word "snuff" is not terribly appealing (and makes it sound like you inhale the product, which is how snuff was originally used), some companies have started calling their moist snuff products "snus' in English-language markets.

Some North American anti-smokeless-tobacco advocates have tried to suggest that the Swedish product is a lot different from the popular U.S. products, using the difference in the names to try to make this point. There are some manufacturing differences between the most popular products in Sweden and in North America. But there is no evidence that the health risks are different. It certainly tells us nothing about a product to know that people who speak different languages have different words for it.