 The terrifying statistic that smoking kills 400,000 Americans each year comes with a silver lining: It isn't actually true. |
A NEWSPAPER AD, full-page. Stark background. Large type. "Cigarettes kill more Americans every year than car wrecks, plane crashes, AIDS, alcohol, drugs, suicides, and homicides combined."
It's a terrifying statistic.
Good thing it's not true.
Of all the falsehoods hammered home in the war on tobacco, none is more pervasive than the one about the body count: 400,000 deaths from smoking per year. Over and over we have been told that smoking is not just unhealthy but massively and catastrophically lethal.
* Elizabeth Whelan, president of the American Council on Health and Science: "If every single day two filled-to-capacity jumbo jets crashed — killing all on board — the death toll would not approach that accounted for each year by cigarette smoking."
* Bill Novelli, founder of the Campaign for Tobacco-Free Kids: "If you want to draw a hierarchy of harms or social problems, you'd probably end up putting tobacco on top. . . . Tobacco is like an atomic bomb on the horizon."
* William Foege, former director of the Centers for Disease Control: "It is quite predictable that in the coming years, the annual global death toll of tobacco will equal the total death toll of the Holocaust in Nazi Germany."
Ghastly as such rhetoric is, worse is on the way. The $246 billion tobacco settlement will be used, in part, to fund even more antismoking propaganda. From print ads, from billboards, from radio and TV, the message will be pounded at you: Tobacco, the worst of all killers, slaughters 400,000 Americans every year.
Except that it doesn't.
"To be blunt," write Robert Levy and Rosalind Marimont, "there is no credible evidence that 400,000 deaths per year — or any number remotely close to 400,000 — are caused by tobacco. . . . The damage from cigarettes is far less than it is made out to be."
Levy is an expert in law and finance who teaches statistics at Georgetown University Law Center. Marimont is a mathematician and scientist who spent 37 years at the National Institute of Standards and Technology and the National Institutes of Health. If you read nothing else about smoking this year, read their article "Lies, Damned Lies, & 400,000 Smoking-Related Deaths" in the new issue of Regulation, a quarterly journal published by the Cato Institute.
What Levy and Marimont call "the granddaddy of all tobacco lies" comes from a 1993 report of the Centers for Disease Control which estimated that 419,000 Americans had died in 1990 of diseases attributable to smoking. A disease was attributed to smoking if the risk of dying from it was greater for smokers than for nonsmokers. But here's the rub — one of several rubs, actually: the CDC included in its death toll diseases for which the relative risk to smokers was statistically insignificant.
It is reasonable to claim that a smoker's lung cancer death was smoking-related, inasmuch as a smoker is 23 times more likely to die of that disease than a nonsmoker. But it is not reasonable to make the same claim for cancer of the pancreas, since the relative risk of that disease for smokers is minuscule — between 1.1 and 1.8 (i.e., a smoker isn't even twice as likely as a nonsmoker to contract pancreatic cancer.) The National Cancer Institute, in guidelines quoted by Levy and Marimont, cautions against relying on relative risks of less than 2. "Such increases may be due to chance, statistical bias, or effects of confounding factors not evident."
Apply that rule to the CDC's 419,000 tobacco-related deaths, and 39 percent of them — 164,000 — disappear. A smoker who dies from pancreatic cancer (or a dozen other diseases on the CDC list with relative risks of less than 2), is not a victim of tobacco. To call him one is to engage in sophistry, not science.
Another statistical trick built into the smoking body count is the failure to correct for other variables. Tobacco use is not the only difference between Americans who smoke and Americans who don't. Levy and Marimont note that smokers tend to be people who also drink too much, exercise too little, eat fewer green vegetables, and have less money. "Each of those factors can be a 'cause' of death from a so-called smoking-related disease," they write, "and each must be statistically controlled for if the impact of a single factor, like smoking, is to be reliably determined."
Controlling for just two of those differences — income and alcohol consumption — reduces the tobacco toll by another 53,000. Take the other factors into account and it would go lower still. But the CDC and the surgeon general treat those factors as irrelevant. "If a smoker who is obese; has a family history of high cholesterol, diabetes, and heart problems; and never exercises dies of a heart attack, the government attributes his death to smoking alone." By that logic, one could just as easily show that 504,000 Americans die yearly from failure to exercise or that 649,000 die from bad nutrition. Like the supposed 400,000 smoking-related victims, say Levy and Marimont, these are "computer-generated phantom deaths, not real deaths."
There's more. For all the talk of protecting children, the average age of death from a smoking-related illness is 72. Measured by years of life lost, smoking is a much smaller problem than alcohol consumption. The number of young people killed by smoking is — zero. All this and more Levy and Marimont calmly explain. Their lucid article provides a fine corrective to the ever more hysterical tone of the antitobacco crusade. You may wish to lay in a copy.
(Jeff Jacoby is a columnist for The Boston Globe).
-- ## --
Follow Jeff Jacoby on X (aka Twitter).
Discuss his columns on Facebook.
Want to read more? Sign up for "Arguable," Jeff Jacoby's free weekly email newsletter.