Dan Werb is an assistant professor of epidemiology at the University of California San Diego and the University of Toronto. He is the author of City of Omens: A Search for the Missing Women of the Borderlands, a finalist for the Governor General’s Literary Award for non-fiction.
As an addictions epidemiologist, I often find myself fielding two basic questions. The first is “So what do you do?” The answer is simple: Epidemiologists study epidemics – their movement across populations and the variables that shape their expansion or retreat. Generally that’s done by creating statistical models that can isolate the influence of one variable on the spread of a particular outcome. The next question is “So how does epidemiology help us understand addiction?” That one is harder to answer.
Epidemiologists build statistical models that map out relationships between variables. Choosing the right variables for a model is where the science of epidemiology starts to shift into an art form. So while there are many approaches to modelling how populations experience addiction, there is no one scientific rule about how to build them. That means that the variables chosen for these models (such as age, gender, income and ethnicity) reflect the personal beliefs of the epidemiologist building them. That means epidemiologic models are susceptible to cultural trends and social change. Over time, we can see them reflecting back to us our own biases, which makes them a reliable barometer of how far society’s understanding of a particular issue has grown. Nowhere is this more evident than in addictions epidemiology, the history of which represents a cautionary tale about how politics, wealth and racism can influence what we think are the root causes of addiction and how we respond to them.
In 1971, in the midst of the Vietnam War and the presidency of Richard Nixon, news began to emerge that roughly one in five U.S. servicemen sent to fight the Viet Cong had become addicted to heroin. Amid a brutalizing war in a country that had become a leading international heroin producer, the addicted GIs – about 30,000 in all – presented a conundrum for the president. Should he let tens of thousands of soldiers linger in a war zone, at risk of succumbing even further to their addictions, or bring them home to the United States, where they might spread the social ill to others?
Before Mr. Nixon’s decision in 1971, scientific work on addiction was mostly conducted by psychiatrists who observed substance use disorders among individual patients. Instead of a population-level issue, substance use was instead largely understood as a personal weakness, the result of a certain “psychopathology” that doomed an unlucky few who were simply not strong enough to handle the rigours of daily life.
Individual psychopathology was a handy theory to explain addiction, largely because it reinforced the structural sexism and racism of the time. For instance, psychiatrists studying addiction in the 1950s observed that African-American populations living in urban centres marked by poverty were disproportionately affected. Nevertheless, these doctors refused to interpret drug use within a social context. To them, it was simply the result of frustration among people incapable of meeting the responsibilities of living in a civilized (read: white Christian male) world.
Even when basic epidemiologic tools such as logistic regression models became more widely used and population-level analyses became the norm, the theory of individual psychopathology wasn’t abandoned. Rather, scientists just mapped it over these new methods to further confirm that addiction was caused by nothing more than basic human weakness. So when clinicians started to consider which variables to put into their statistical models to explain why some people became addicted to drugs and others didn’t, the focus was almost entirely on group personality traits, which essentially can’t be changed. As one seminal study from 1962 put it, “the addicts seen in American treatment facilities are predominantly young, male, psychopathic, immature individuals drawn from foci spotted about certain big city slums … The relationship to the so-called psychopathic personality is close and several epidemiologic characteristics of psychopathology and addiction coincide fairly well.”
The reference to “big city slums” here was, of course, racist dog-whistling. And what about that throwaway line about psychopathology and addiction being often indistinguishable? For the authors, drug-addicted people were “suffering from strongly passive-dependent needs … Their personality is also called immature and pseudo-aggressive since the aggression when it does occur is self-defeating and ineffectual … the self-destructive effect of the behavior of the addict is similar to that of the classic psychopath.” The takeaway is that “addicts” – like psychopaths – are born and not made. The inherent weakness of the person was beyond medical care. This psychopath argument about addiction was the kind of scientific theory that we should always distrust: one that somehow manages to explain away a societal ill while exonerating the society itself. And perhaps it would have remained the prevailing view if it hadn’t been for one little problem: the behaviours of the doctors themselves.
While up until the 1960s the medical and scientific establishments were united in the view that addiction arose almost solely from an immutable personality defect, interest in the emerging scientific field of epidemiology had motivated investment in building large and long-term observational cohort studies. Although incredibly useful in showing that specific lifestyle factors cause certain health outcomes, recruiting participants into these kinds of studies can prove challenging. To get around that, many early epidemiologic investigations originally relied on physicians as volunteer participants, given that they were easy to follow and generally willing to participate.
So, in the 1950s and 1960s, a glut of studies following physicians were launched in the United States. At the time, the main concern among epidemiologists was that these studies’ findings were likely biased because physicians generally exhibited higher-than-average levels of overall health and wealth, making them poor representatives of the overall population. What nobody counted on was that in the area of drug addiction, these studies would reveal an uncomfortable truth: It turned out that doctors were disproportionately likely to become addicted to drugs and alcohol. This finding was the tip of the spear that forced the scientific community to lurch beyond its origin story of addiction as personal weakness or insanity.
With physicians more likely to become addicted to drugs, compared with the general population, it became a lot more difficult to argue that a drug-dependent person was a “classic psychopath” or inherently “immature and pseudo-aggressive.” The situation was particularly untenable given that, during the fifties and sixties, physicians were the people running most epidemiologic studies and authoring the scientific manuscripts about drug use. They were, unsurprisingly, loath to suggest that the high prevalence of drug addiction among members of their vocation was caused by the fact that doctors are all psychopaths.
And so, instead of blaming that same collective form of psychopathology that they had diagnosed as innate to African-Americans, Latinos and women, epidemiologic papers about addicted doctors quietly gravitated toward different language to talk about drug use and its effects.
In one study from 1966 that compared 100 physicians treated for addiction with 100 matched controls, the authors – physicians themselves, of course – wrote, with a level of subtlety absent in studies of drug use among black Americans, that they found “no correlation between psychiatric diagnosis and drug used” and the study’s participants. As far as the researchers were concerned, doctors couldn’t be crazy, even the ones that overindulged. In a lingering sign of the times, though, the factors the authors deemed most likely to increase the risk of drug use reflected myopic ideas about the root causes of addiction. These included whether participants were married, whether they were Protestant and whether they came from the American South.
Another study, published in The New England Journal of Medicine in 1970, reported that after 20 years of following a group of college students, half of whom had gone into medicine, twice as many of the physicians had used drugs as the group of people who, one assumes, found less respectable careers. Here, the authors again included variables they assumed most relevant to addiction: having had a feeding problem in infancy, having had a private-school education and scoring badly on a math test. Today this kind of paper wouldn’t even make it to a scientific journal editor’s desk, let alone get published.
What these mid-century epidemiologists overlooked about substance use among doctors were the high levels of stress, anxiety and lack of sleep that characterize the medical profession. Coupled with ready access to highly addictive pharmaceutical drugs and a culture of intense competition, doctors were primed to self-medicate.
Having pragmatically turned themselves into their own guinea pigs, doctors had inadvertently revealed their own heightened drug use and, with it, the fatal flaw behind the racist and sexist addiction science they had popularized. This led to only one conclusion: If morally upstanding, intellectually sophisticated white men were succumbing to addiction in droves, then it could not be a disease of the mind. The upshot was that the kinds of variables included in addiction models expanded beyond an individual’s personality or upbringing. While this was a welcome step forward in our understanding of addiction, it came on the back of the unsettling cultural truth that when it happens to wealthy white guys, it isn’t their fault.
Meanwhile, the Vietnam War raged and Mr. Nixon found himself with 30,000 opioid-dependent GIs waiting overseas. Like the drug-dependent physicians, the situation of the soldiers upended the psychopathology theory of addiction. In its place, and reflecting the growing power of epidemiology to explain a vast array of health and social phenomena, an increasing number of thinkers began to describe addiction using language from this emergent discipline. In 1971, the same year that Mr. Nixon would make his decision about the fate of the GIs, the American Medical Association published a report describing drug dependence as arising from the interplay of three factors: the agent (or pathogen), the host and the environment, otherwise known as the epidemiologic triangle. Whereas before physicians running addiction studies had focused exclusively on the host – that is, those “weak” groups in the U.S. inherently susceptible to drugs – this new approach better explained how drug use spread across all manner of populations.
Mr. Nixon’s personal views about drug use were cut and dry. “Dope? Do you think the Russians allow dope? Hell no,” he is recorded as saying in the White House in May, 1971. Despite the intensity of his anti-drug feelings, he was at a loss to find a solution to the addicted GI issue that was consistent with his own thinking.
Eventually, he relented in the face of the inevitable. With scientific studies increasingly touting the benefits of using methadone – an opioid itself – to treat heroin addiction, he was left with little choice but to base his policy on evidence. So, despite launching the U.S. war on drugs in June, 1971, Mr. Nixon also agreed to provide all 30,000 opioid-addicted GIs in Vietnam with methadone-assisted treatment, instructing military personnel to only allow soldiers to board planes for the United States if they were no longer exhibiting symptoms of addiction. This quarantining of “infectious” cases is an old strategy to contain epidemics. While levels of substance use disorders among U.S. military veterans remain higher than the national average, Mr. Nixon’s approach averted what the White House had feared would be a catastrophic expansion of heroin addiction across the United States. At the highest levels, the epidemiologic triangle – agent, host and environment – had managed to persuade a president hell-bent on punishing addicted people to treat them instead.
Nowadays, the statistical models epidemiologists build to explain health do more than account for feeding problems in infancy. Like good art, good statistical models reflect the world as we know it and illuminate it further. The tenets of personal responsibility, prosperity and an obligation to avoid harm that make up the prevailing American ethos have been eclipsed by the reality that systemic racism, a self-protecting oligarchy and political structures resistant to change (and willing to punish those advocating for it) limit the capacity of some among us to avoid injury or ruin. We know it matters less whether someone could take the nipple as a baby as whether they were born into poverty. And so we choose one model and not the other.