We Should Celebrate,Not Censor,Learning From Epidemiologic History

Background

Chronic traumatic encephalopathy (CTE) is an occupational neurodegenerative disease ¹ with widespread and grave consequences for individuals and society. Unfortunately, the Occupational Safety and Health Administration (OSHA), my former agency, does not regard it as a priority, and many medical professionals and others continue to express more doubt about it than the evidence warrants. CTE is newsworthy for its prevalence in former contact-sport athletes—in American football, rugby, soccer, and other sports—but there are many blue-collar occupations with repeated head impact (RHI) whose workers are therefore also at risk. These occupations include firefighting, military service, logging and tree-trimming, some construction and warehousing jobs, and long-haul commercial driving. ^a

Any consensus among scientists or among the general public that RHI does or does not cause CTE, or increases the risk of CTE, has enormous financial stakes attached. The National Football League (NFL) alone reaps about $20 billion annually in revenue, and has already paid out over $1.3 billion to former athletes who have suffered concussions, though the settlement does not compensate for new cases of CTE per se. ^b The intense debate about causality here is very reminiscent of the decades of scientific controversy that surrounded one of the few products with revenue exceeding that of professional sports: tobacco (with annual revenues of nearly $90 billion in the United States).

I was privileged to be a coauthor of a comprehensive analysis ² of the epidemiologic, toxicologic, and mechanistic evidence that bears on whether RHI can cause CTE and related latent neurological diseases. We applied the Bradford Hill criteria ^c and concluded that all 9 criteria were met, most amply so. Shortly after that article was published, the National Institute of Neurological Disorders and Stroke changed its position on the RHI → CTE relationship from “CTE is associated with RHI” to “CTE is caused in part by repeated traumatic brain injuries.” ^d

But despite this ample and mounting evidence base, a wide variety of arguments are still raised and repeated that the face-value evidence should be ignored, or that it suffers from confounding, various types of bias, and/or failure to account for unmeasured genetic factor(s) that purportedly could explain the CTE cases without RHI as a cause. Recently, an article was published claiming that the skepticism raised about the RHI → CTE relationship is not analogous to the doubt manufactured by the tobacco industry during the last century about the smoking → cancer relationship. ³ The apt and inapt features of this analogy are the main focus of this article.

I’ve coauthored several articles on various logical fallacies in the arguments that minimize or doubt the seriousness of RHI.^4,5 In particular, observations that there exist “people with RHI who have not developed CTE,” ^e and “people with CTE who have no history of RHI” ^f show only a lack of understanding of how public-health evidence works: both observations are completely unremarkable (indeed, trivial) and cast zero doubt per se on the RHI → CTE causal relationship. The presence of exposure without disease in some individuals is, of course, exactly the definition of risk itself: a probability of harm between 0 and 1, which some individuals will never suffer despite the risk. Surely the statement “I’ve heard of someone who smoked for 50 years and died without lung cancer” would be ignored or derided if offered to exculpate smoking as a cause of lung cancer: smoking causes lung cancer in many, but not all. Similarly, “I’ve heard of someone with lung cancer who never smoked” merely reflects the fact that other agents (e.g., indoor radon) can cause this disease, and in no way casts doubt on what smoking can cause. Related fallacious arguments are often made with respect to the symptoms of CTE (cognitive decline, severe mood changes): anecdotes about individuals who were found with CTE lesions but no symptoms, or ones who had these symptoms but no CTE lesions, are again completely to be expected in a disease that can be indolent, and for symptoms that can easily be caused by other conditions.

More superficially reasonable objections to the validity of a causal RHI → CTE relationship are also frequently offered, generally accusing epidemiologic studies of workers and others with and without RHI and/or CTE of being flawed. A detailed showing of how one or more positive studies were, say, tainted by an unmeasured confounder, recall bias, and/or selection bias might lead to a legitimate reconsideration of the overall strength of the evidence. However, without evidence, simply invoking jargon related to these valid concepts is not compelling on its own. ^g Many of the most vocal skeptics seem oblivious to the fact that a true confounder must be associated both with the adverse outcome and with the exposure of interest—so, for example, it makes little sense to offer up steroid abuse ⁶ as an alternative cause to RHI, without acknowledging that CTE rates are far lower in professional baseball than in professional football, despite the historically widespread abuse of steroids in the former sport.

This clash between face-value evidence and skepticism should be very familiar to readers of NEW SOLUTIONS, who have experienced or read about waves of “manufactured doubt” about virtually every important cause of occupational or environmental disease.^7,8 In short, the first wave, roughly speaking, was the hiding/denying of data about lead, asbestos, radiation, and other toxicants. Then, the modus operandi shifted to explaining away the data—invoking “mice are not little men”^9,10 to pooh-pooh toxicology findings, or various forms of “Fisher's Fallacy” ^h to divert attention from significant relative risks found via epidemiology. More recently, the watchword has become “sophistication.” Examples of “sophistication” demanded by skeptics of the RHI → CTE link include:

complicated models of dose-response (including hormesis, an asserted beneficial effect at low doses of an exposure); and

Recently, the Journal of Science and Medicine in Sport (JSAMS) published a commentary written by 3 sociologists from the United Kingdom,³ who argued that no one should use the “tobacco analogy” to criticize the arguments against a link between RHI and CTE. Malcolm et al. concluded that “the time for politically motivated analogies has now passed,” arguing that head impact is unlike tobacco in that it is not addictive, that in contrast to the 1 billion people who smoke worldwide, “only a tiny proportion of the world's population will ever become elite athletes,” and that while tobacco is “a single causal factor” causing a single primary health outcome (lung cancer), head impact involves a “heterogeneous group of conditions and mechanisms.”

I submitted the following reply to JSAMS, ⁱ and the editor-in-chief rejected it, stating that it would have to be “toned down to reach (the) status of serious exchange among scientists by use of appropriate language.” I believe that “toning it down” would not have been appropriate because of the importance of manufactured doubt as a real phenomenon and as a weapon against regulatory and other solutions. Profoundly harmful recommendations can be couched in “civil and serious” language and insisting on “civility” can be a ploy to avoid criticism and gaslight the critic.

Response to Malcolm et al. (Rejected by the Journal of Science and Medicine in Sport)

Malcolm et al. decry the continued use of “the tobacco analogy” in discussions of the epidemiology and policy of sport-related RHIs and CTE. If referencing the needless death toll from many decades of “manufactured doubt” sowed by tobacco lobbyists and their hired scientists makes some RHI researchers and sports executives defensive and uncooperative, I can appreciate the sensitivity about the facet of the analogy that involves “advocacy for hire.” But as an epidemiologist and former (U.S.) government regulator of occupational health and safety, I hasten to add that the tobacco analogy is actually quite germane to explain two scientific concepts that these authors may not be well-versed in: (1) dose–response relationships; and (2) the role of “complexity” in quantitative risk assessments.

At the outset, it's important to recognize a few of the “straw men” Malcolm et al. use to trivialize the analogy. No one to my knowledge has claimed that cigarettes and contact sports are similar in “addictive” properties, or that the burden of mortality and morbidity from contact sports is anywhere near as large as that from tobacco. Analogies, as the authors understand, are “as much about invoking comparison as they are about precision.” So, two hazards can be very instructively alike along some important dimensions and dissimilar along others—after all, we teach high school students that the human circulatory system and a tree both have fractal characteristics (which is to say they branch out into progressively more narrow units), but we don't teach them that humans are made of wood.

And the fact that contact sports have physical and psychological benefits, unlike cigarettes, is another complaint Malcolm et al. make that actually favors the analogy. For decades, ¹¹ the cigarette industry and their favorite physicians touted the supposed benefits of that product. I do not suggest that we will learn that the benefits of contact sports are also illusory. Instead, contact-sports enthusiasts offer only a false choice: that young athletes need to accept risks of grave disease in order to obtain the benefits of physical activity and camaraderie. There are, of course, thousands of ways to obtain these important benefits without the risks—team sports that don't depend on relentless RHI, individual sports like tennis or weightlifting, etc. When NFL Commissioner Roger Goodell told reporters ¹² that “There's risk in everything. There's risk in sitting on the couch,” scientists who understand elementary logic (and its misuse) could well have imagined an advertisement from a fictitious “World Russian Roulette Association,” proudly stating that “It Gets Your Heart Rate Up.”

But it is in the area of dose–response analysis where the tobacco analogy shows how little some CTE skeptics have learned from history. The groundbreaking epidemiologic research on tobacco that began in the 1950s ¹³ further showed that “Haber's Law” (risk is a function of concentration multiplied by time) applies to most environmental toxicants, which is why we now generally measure tobacco exposure in “pack-years” and exposures to solvents in ppm-years. ¹⁴ When this work began, mainstream science already understood that “high” doses of toxicants could cause rapid and obvious harm—say, instantaneous rupture of the human eardrum at noise levels exceeding about 150 decibels (dB). Only through painstaking work on tobacco, asbestos, benzene, and many other substances did we learn that with enough repetition, very “low” doses can cause similar or equally serious harm, such as occupational deafness from long-term exposure to 90 dB (a sound pressure one one-millionth as powerful as 150 dB). No sensible person would be surprised to hear that one lone night of binge drinking, while not good for you, cannot cause cirrhosis of the liver—but that 3 drinks per day, every day, for a decade, likely will.

And yet, deniers of the RHI → CTE relationship are trying to appear “reasonable” by reluctantly agreeing that a few traumatic brain injuries can cause latent disease in an athlete, while still expressing surprise or derision that thousands of nonconcussive impacts could possibly cause harm. The history of environmental health science, especially tobacco, shows how foolish such a bifurcation is.

The authors also repeat the claims that the hazards of tobacco are straightforward while those of RHI are “multifaceted,” and that (only) in the latter area must we discover how “sex, genetics, medical history, and environmental and lifestyle factors” cause “some people to develop CTE and others not.” ³ These misunderstandings instead show clearly why tobacco and RHIs are usefully analogous.² Cigarette smoke is a complex mixture of dozens of toxicants, and it causes many health effects other than lung cancer. More fundamentally, we still have no idea—and do not need to have one!—why some smokers get lung cancer and others do not. This is what “risk” is—a probability like any other “roll of the dice”—and why at the individual level, a person at lower risk can succumb while a person at higher risk skates by, but why at the population level, response is proportional to dose.

The overwhelming message of the “tobacco wars,” other than that revenue can derange logic, is that we did manage to warn and control and reverse the time trend of smoking-related cancer even though we didn't and still don't have virtually any idea how genetics, lifestyle, randomness, and molecular events interact to initiate, promote, and proliferate malignancies in some smokers.^{15, j} Thank goodness we did not forever allow true but desperate claims that “the science is still in its infancy”^{16, k} to lead us to treat tobacco as a mysterious and possibly benign exposure, or one that “only affects people with bad genes.” The science of either hazard may always be in its infancy, but we need to be grown-up about our obligation to treat “probable” health hazards with honest disclosures and low-regret controls. Furthermore, we need to keep one eye on an eerie parallel between e-cigarettes and contact sport: both industries target children with advertising. ¹⁷

So, to the authors’ conclusion that “the time for politically motivated analogies has now passed,” I would add that declaring an analogy verboten can also be politically motivated. Attempts to ignore the lessons of past hubris, and to substitute static for signal, concern me more than the possible overuse of an analogy.

Conclusions (material not originally submitted to JSAMS)

The obvious embarrassment that some minimizers of RHI feel when confronted with the “tobacco analogy” is instructive for other occupational and environmental controversies in two complementary respects. First, it is always to the benefit of skeptics to highlight the obvious—that we can never “know” with complete certainty that exposure to X can cause disease Y—by sowing doubt about X and Y and/or by exaggerating the extent to which we actually “know” that some other exposure causes some other disease. It is for the latter purpose that Malcolm et al.—like many others—try to drive a wedge between what we “know” about tobacco and what we know and need to know about RHI and latent brain disease.

But the larger problem with doubt is that advocates use it to pollute the needed parallel relationship between the degree of certainty and the strength of action. Striving for total certainty is noble, but refusing to consider any action in the absence of total certainty is myopic at best, and a deadly hoax at worst. By implying that “until we know everything, we can't do anything,” purveyors of doubt seek to thwart any discussion of the obvious and opposite tenet: “as we learn more, we should do more.” As we imagine moving from darkness to intense light along a spectrum of evidence, we should contemplate moving from utter inaction to, for example, honest disclosures of risk and uncertainty to potential victims, to simple, low-regret actions, ^l to more expensive controls, to strict regulation, to bans—all in proportion to what we learned.¹ We should also be willing to move “backwards” on this spectrum should we learn something exculpatory. ^m As an expert witness in many court cases where manufacturers have failed to warn downstream customers (as required by the OSHA Hazard Communication Standard, among other regulations and norms), I regard honest disclosure as one of the easiest and most useful actions, long before the elusive “known causality” has been reached. Fixating on what we (purportedly…) don't know is thus an artifice to avoid focusing on what we can do. We can disclose, we can control, and we can conduct research, as we certainly did with tobacco.

Specifically, I and many others have recommended various low-regret actions that the responsible institutions (particularly the NFL and the National Collegiate Athletic Association [NCAA]) could undertake, perhaps in partnership with OSHA or NIOSH. “Low-regret” actions to reduce an uncertain risk ¹⁸ are steps that will clearly have benefits that justify their cost if the harm turns out not to be caused by the exposure being reduced, and that may even have “negative costs” regardless. An example of an action with a “negative cost” is the 2011 decision by the NFL to move the kickoff point forward to the 35-yard line. Moving the kickoff point not only reduced the number of concussions during kickoffs, but also reduced the number of fractures, ligament/tendon injuries, and the like. ¹⁹ In addition to the suggestions already mentioned in note l above, here are 4 different categories of interventions that might have substantial net benefit especially if (but even if not) we become even more confident that RHI in contact sports causes a substantial risk of CTE and other neurodegenerative diseases:

Rule changes: In addition to changes the NFL has already made (changing the distance of the kickoff to reduce the number of runbacks where all 22 players accelerate to top speed before hitting and tackling), I’d suggest to the NFL and NCAA that they consider the example of the National Basketball Association (NBA) with respect to how dangerous conduct is more effectively deterred. In football, there are penalties for spearing (using one's helmet as a weapon and the first point of contact), grabbing the opponent's facemask, and the like, but overwhelmingly, these penalties result in a single loss of yardage, and occasionally a monetary fine levied against the offending player. In contrast, a team will only have a strong incentive to insist that its players don't commit these kinds of fouls if it's possible that a player will be ejected from a game, affecting the all-important won/loss result. In 2023, out of more than 43,000 separate plays during NFL games, there were only 12 instances where a player was ejected for a flagrantly dangerous foul. ⁿ The NBA, in contrast, has criteria for automatic ejection but also has a system where a player's less serious but still worrisome (“flagrant 1” and “technical”) fouls are tallied, and he can be suspended for one or more games for exceeding a given number of these offenses over the course of a season. Any policy changes that would truly deter the kind of conduct the NFL and NCAA claim to want to minimize would signal a more serious commitment to making concussions and nonconcussive head impacts less frequent and severe. It would also signal a more serious commitment if the NFL would define as “unsportsmanlike conduct” (see page 50 of the current NFL rulebook) ²⁰ meriting possible ejection not merely actions like taunting or touching an official, but unsafe on-field maneuvers that risk significant head impact to self or others.

The true relevance of the tobacco analogy to CTE and to many other occupational/environmental diseases is that it reminds us that, however slowly society reacted to the disquieting information coming from epidemiologic, toxicologic, and mechanistic work, we did progress in tobacco policy from “nudges” to warnings to economic disincentives to narrow controls (airplanes, hospitals) to broad-based controls. It is ultimately disappointing that the advocates complaining most bitterly about how analogies “distract” us from progress are contributing so little to progress themselves.