When Did Smoking Become an Addiction?

Stephan Risi

When did smoking become an addiction? Was it when anecdotal evidence connected smoking to addiction? Or when tobacco companies started to gather evidence that smoking was addictive? Or was it, finally, when the link between smoking and addiction became widely known? The first two possibilities come with clear answers. If we’re interested in anecdotal evidence, we can point for example to letter that John Quincy Adams wrote in 1845, where he noted that in his early youth, he was “addicted to the use of Tobacco in two of its mysteries, smoking and chewing.” John Quincy Adams. "Letter, Re: Addiction to Tobacco, John Quincy Adams." 19 Aug 1845. Or, if we’re interested in scientific research, we can point to the late 1950s and early 1960s as the important juncture, when researchers at British American Tobacco studied nicotine extensively and found, in the words of their chief scientist that “smoking is a habit of addiction.” Charles Ellis. "The Smoking and Health Problem." 29 Nov 1962.

However, it is the final question—when did the link between smoking and addiction become widely known—that is of most importance. Only when this link became popularized can smokers be expected to have heard about it and only then should they be held legally accountable. Only then did it reshape understandings of who is responsible for excessive smoking from the individual’s “weak will” to tobacco’s main addictive ingredient, nicotine. And only then did new quitting tools like nicotine gums and patches gain footing.

This last question is harder to tackle than the first two. Answering it requires less individual documents and rather statistical averages: when do we find the term “nicotine addiction” not just in a few individual documents but in large groups of documents? And which other terms rise alongside or are correlated with it? The goal of Tobacco Analytics is precisely to provide tools to summarize and visualize the contents of all the documents in the TTDL and to thereby make it easier to tackle big, fuzzy questions.

This essay is intended as both a case study and introduction to the possibilities offered by the frequency charts of Tobacco Analytics. On the one hand, I will present evidence that points to the late 1980s and early 1990s as the time when the link between smoking and addiction became widely known and started to change smoking patterns, treatment modalities, or understandings of responsibility. And on the other hand, I will show how to use the frequency charts and what its potential pitfalls are.

Strange Spikes and Other Curious Patters:
Some Notes on the Tobacco Documents

As with any visualization platform, it's tempting to just jump in and start charting. However, there are two pitfalls to be aware of:

  • Only a small number of documents dates before 1950, which means that single misdated documents can cause huge spikes in the early 1900s.
  • Most documents before 1998 come from the internal archives of tobacco companies. After 1998, the majority are litigation documents like depositions or trial testimony. Because of these two factors, we can often find strange patterns before 1940 or after 1998.

The frequency chart of "addiction" demonstrates both of these patterns. It seems to suggest that addiction appears most often in the early 1900s and after 2005. However, those spikes are largely an artifact of misdated documents and changing document collections.

Misdated documents are usually to blame for unexplained spikes in the early 1900s. Only 73.000 of the over 14 million documents in the TTDL date from before 1950. Hence, when a single document from the 1980s or '90s gets mistakenly assigned a document date in the early 1900s, it can cause a large frequency spike simply because there are very few documents for any given year. That's exactly what happened here: A compilation of individual documents from the late 1980s was assigned the year 1912 as a document date. [Author Files regarding the U.S. Surgeon General Report on Addiction] Since this document contains the word "addiction" more than 200 times, it is solely responsible for the spike in 1912. The easiest way to identify these documents is to use the text passages search to identify the documents responsible for a spike.

In contrast, spikes after 1998 are usually attributable to changes in the kinds of documents available in the TTDL. The vast majority of the TTDL documents became publicly available through a lawsuit between the Attorneys General of 46 states and the five largest U.S. tobacco companies. The result of this case was the 1998 Master Settlement Agreement, which, in addition to annual penalties, forced the tobacco companies to make all the documents disclosed during the discovery phase of the case publicly available. The MSA documents make up the bulk of the TTDL archive and it is these documents that have given us unprecedented insights into the inner workings of the tobacco industry. After 1998, however, the documents in the TTDL change.

We can use counts of the word "the" as a proxy to gain a sense of the distribution of documents across collections and years. What this graph makes clear is that the number of documents swiftly declines after the 1998 MSA and becomes more litigation focused: On the hand, a larger proportion of documents stem from the "Depositions and Trial Testimony" collection, which aggregates litigation documents. On the other hand, most Philip Morris documents after 1998 are part of a 2006 RICO case against Philip Morris.

To circumvent both the sparsity of documents in the early 1900s and the change in document collections after 1998, the frequency charts use 1940 to 1998 as the default time frame. You can, however, modify this by moving the slider on the left.

From "Smoking Habit" to "Nicotine Addiction"

To find out when smoking became linked to addiction, we can search more specifically for “nicotine addiction, smoking addiction, tobacco addiction”. Tobacco-analytics allows you to search for multiple terms—they just need to be separated by commas. It also permits you to search for up to 5-grams, that is: 5 words that form an expression like ‘cigarette smoking causes lung cancer’.

Here, a much clearer picture emerges: ‘tobacco addiction’ and ‘smoking addiction’ barely appear. The key term is clearly ‘nicotine addiction’. However, it hardly shows up until the mid-1980s and peaks—by number of appearances—in the 1990s.

There's an important distinction to be made here between the knowledge of tobacco industry scientists and the knowledge of average smokers. Most tobacco companies figured out that nicotine was ultimately the reason why people smoked in the 1960s and early 1970s. Claude Teague In 1972 Claude E. Teague, Director of Corporate Research at R.J. Reynolds, makers of Camel cigarettes, drafted a confidential memo about “the crucial role of nicotine” in the cigarette business that summarizes the industry's understanding in the early 1970s:

In a sense, the tobacco industry may be thought of as being a specialized, highly ritualized and stylized segment of the pharmaceutical industry. Tobacco products, uniquely, contain and deliver nicotine, a potent drug with a variety of physiological effects. [...] Our Industry is then based upon design, manufacture and sale of attractive dosage forms of nicotine, and our Company's position in our Industry is determined by our ability to produce dosage forms of nicotine which have more overall value, tangible or intangible, to the consumer than those of our competitors.

However, even within the industry, few were prepared to discuss smoking as an addiction to nicotine, in large parts because of the potential regulation that this position might entail. William Dunn, Teague's counterpart at Philip Morris, for example, cautioned against using a pharmacological model: "[D]o we really want to tout cigarette smoke as a drug? It is, of course, but there are dangerous FDA implications to having such conceptualizations go beyond these walls." William Dunn, "Memo to Dr. Wakeham from W.L. Dunn about Jet's Mony Offer", February 19, 1969. Hence, while the discovery that nicotine was addictive had an immense impact on the industry, it's hard to discern its impact through the trajectories of terms like "nicotine addiction" because the industry's scientists had good reasons to avoid discussing nicotine addiction openly and often opted for more innocuous code words like "satisfaction."

Hence, outside the small circle of industry scientists and executives, smoking and and addiction were rarely linked before the 1980s. Instead, smokers might have referred to their "smoking habit."

The difference between habit and addiction is important. From 1950 to 1964, the WHO classified drugs as “addicting” and “habituating”. Expert Committee on Drugs Liable to Produce Addiction, Report on the Second Session, Geneva, 9-14 January 1950, vol. 21, World Health Organization Technical Report Series (Geneva: World Health Organization, 1950). Drugs were considered to be addicting if they led to compulsive use and chronic intoxication. Habituation, in contrast, was defined through the desire but not the compulsion to take a drug. The label had important effects: addicting drugs underlay international control and trade restriction whereas habituating drugs only needed to be properly labeled so as to warn the customer of the potentially habituating nature of the drug. For this reason, the decision about whether a drug was addictive, habituating, or neither was ruled primarily by financial interests: the pharma industry made sure that cash cows like amphetamines and barbiturates did not fall into the “addicting” category despite fulfilling all of the requirements. Nicolas Rasmussen, "Maurice Seevers, the Stimulants and the Political Economy of Addiction in American Biomedicine," BioSocieties 5, no. 1 (2010).

Maurice Seevers, himself a consultant to American Tobacco, used the same distinction between “habituating” and “addicting” in the 1964 Surgeon General's Report to argue that smoking was a habit, not an addiction. Of course, as an expert on the topic he must have known that the WHO would abandon the tenuous distinction between habit and addiction in the very same year and unify them under the label “dependence”. WHO Expert Committee On Addiction-Producing Drugs, Thirteenth Report, vol. 273, World Health Organization Technical Report Series (Geneva: World Health Organization, 1964). However, despite the fact that the WHO abandoned habit and addiction, the effects of the distinction lingered. “Habit” linked smoking to lesser evils like drinking tea or coffee. Sure, too much of it might be bad for you, but too much usually meant more than one pack per day and it was assumed that you could always stop if you wanted. Addiction, in contrast, linked smoking to hard drugs and substance abuse.

This raises the question: why did smoking turn from a habit into an addiction in the late 1980s? My thesis is that during the 1980s, smoking became biologized—it was transformed from psychological and social process into a biological one. Through this transformation—the inscription of smoking into the brain—it became an addiction because scientists could show that it affected the same neural pathways as heroin and cocaine. Earlier theories held that people smoke because their personalities predisposed them or because it provided relief from stress. In the 1980s, the focus shifted and narrowed from the general psychological effects to smoking to the very specific action of one molecule—nicotine—on the brain.

A New Paradigm Established: Addiction in Neuroscience,
Smoking Cessation, and Tobacco Litigation

The key change was the rise of neuroscience in the study of nicotine, which allowed researchers to compare the effects of nicotine to those of drugs like heroin and cocaine. We can trace this rise through a Boolean search, by searching for “nicotine*”. This returns the 10 most frequent terms starting with “neuro” like “neuron” or “neurology”. To aggregate the individual trajectories, the terms are displayed in “stacked” mode—one on top of the other, which makes general trend that underlies all these terms more visible.

If we want to know when neuroscience entered not just the tobacco documents in general, but more specifically the study of nicotine, we can enter "nicotine" as a term filter. Using the term filter "nicotine" means that we're not using all the tobacco documents but instead base our analysis on only the 100 word passages that contain the word "nicotine." That is: We split all tobacco documents into 100 word chunks and only keep those that contain the word "nicotine." Then, we look when neuroscientific terms appear in those selected passages. This gives us a way to figure out when nicotine and neuro terms start to come together. Term filters are useful because we don't need to define how nicotine and neuro are connected in specific expressions (e.g. "nicotine addiction is a neurological disorder"). Instead the term filter gives us all passages where nicotine and neuro appear together.

Using the "nicotine term filter, we find a steady rise throughout the 1980s. The peak almost coincides with the 1988 Surgeon General's Report, which summarized recent findings and concluded that “the processes that determine tobacco addiction are similar to those that determine addiction to other drugs such as heroin and cocaine.” U.S. Department of Health and Human Services, The Health Consequences of Smoking : Nicotine Addiction. A Report of the Surgeon General (Washington, DC: GPO, 1988), I. Neuroscience made that claim possible: if nicotine, heroin, and cocaine all affect the same pathways in the brain, then they are all addictive.

This neuroscientific approach replaced older understandings of smoking based on personality types and psychological responses. Indeed, we can find both “psychology” and “personality” dropping precipitously in the 1980s, just like “smoking habit”.

Correlation, of course, doesn't imply causation. In this case, however, the link is quite clear. The tobacco industry had for decades funded the work of psychologists who argued that smokers reached for cigarettes either because their because of stress or because their personalities predisposed them. Hans Selye is a good example of the former, psychological model, arguing that smoking could act as a defensive mechanism against the stresses of modern life. Mark Petticrew and Kelley Lee, "The 'Father of Stress' Meets 'Big Tobacco': Hans Selye and the Tobacco Industry, American Journal of Public Health 101, no. 3 (March 1, 2011): 411-418. Hans Eysenck, in contrast, was a proponent of the personality model, claiming that smokers were usually extraverted "sensation seekers". As a stimulant, he saw nicotine was a natural fit for this group, though without giving much thought to the possibility that nicotine could also produce addiction. Hans Eysenck, "Personality and the Maintenance of the Smoking Habit, in W.L. Dunn (ed.) Smoking Behavior: Motives and Incentives, Oxford: V.H. Winston, 113-146.

With this new emphasis specifically on the biological effects of nicotine over the psychological effects of smoking in general came a host of new treatments: nicotine gums, patches, inhalers, and sprays. All of them are predicated on the idea that smoking can be reduced to nicotine administration and that replacing the nicotine is a primary requirement for smoking cessation. Without a novel paradigm that understood smoking as an addiction to nicotine, they would simply not have made sense.

At the same time, addiction entered into litigation. To explore this link, we can look when "nicotine addiction" appeared in the "Depositions and Trial Testimony (DATTA)" collection, which aggregates litigation documents like depositions, testimonies, and exhibits.

Here, we find a familiar pattern: nicotine addiction becomes prominent in the late 1980s through a succession of spikes, which correspond to different waves of litigation. The largest of these spikes, from 2008 to the present, is the result of the ongoing Engle progeny cases in Florida. the largest of which were the still ongoing Engle progeny cases in Florida. The new scientific consensus about nicotine’s addictive properties undermined one of the tobacco industry’s strongest arguments: until the 1990s, industry lawyers argued successfully that even if smoking was harmful, only the smoker should be held responsible if she became ill or died because she had freely chosen to smoke. Now, plaintiffs could rely on scientific evidence to argue that smokers were at the mercy of a powerfully addictive substance, comparable in this respect to heroin or cocaine. Given these new facts, who could argue that smokers were completely free in their choice to smoke?

In conclusion, the biologiziation of smoking, the development of new cessation products, and the entry of nicotine addiction into litigation all worked together and fueled each other. Together, they established and promoted the idea that smoking is addictive to the point where it became widely known in the late 1980s and early 1990s. This moment of wide knowledge is far more relevant than both early anecdotes (going back to the 19th century) or secret tobacco industry studies of nicotine (going back to the late 1950s) because only when the link between smoking and addiction became popular did it affect policies, treatment, or litigation.

However, tracing broad shifts such as this one requires looking at the LTDL archives from 30.000 feet. From this height, we may lose many subtle differentiations but it enables us to survey the whole landscape and to either reach new conclusions or find support for old hunches. The frequency charts cannot replace ground-level work with individual documents. But they can support this work and help dating slow, gradual changes.