February 16, 2012

Why Do Public Health Advocates Lie About the Risks of Smoking?
August 30, 2002


Joseph Bast is president and CEO of The Heartland Institute, a 28-year-old national nonprofit... (read full bio)

At the Northwestern train station in downtown Chicago, commuters are met with a billboard that reads: “Odds of dying in a car crash: 6,200 - 1; dying from smoking: 3 - 1.”
The ad, paid for by the Illinois and Chicago departments of public health, is obviously meant to encourage smokers to quit. But is it accurate?

Apples and Oranges
The billboard’s message is fraudulent two ways. First, it compares the risk of driving a car for one year with the risk of smoking for a lifetime, an apples and oranges comparison. The “odds of dying in a car crash” were apparently derived by dividing the total population of the U.S. by the annual number of vehicular deaths in the U.S.: about 280 million divided by 45,000. Given a life span of 75 years, the lifetime odds of dying from a car accident are really about 90 to 1, not 6,200 to 1.
Second, the billboard exaggerates the odds of dying from smoking. The actual odds a smoker faces of dying from smoking before the age of 75 are about 12 to 1. Describing how this estimate is derived simultaneously rebuts much higher estimates popular in the public health community.
The Surgeon General, using statistical models, estimates a total mortality risk from smoking between .18 and .36. The billboard’s claim of “3 - 1,” or .33, is at the high end of this range.
But the basis for these numbers has come under heavy fire from experts, such as Robert Levy, an adjunct law professor at Georgetown University, and Rosalind Marimont, a mathematician and scientist who worked for the National Institutes of Health for 16 years before retiring.

Lies and Statistics
In a 1998 article titled “Lies, Damned Lies, and 400,000 Smoking-Related Deaths,” Levy and Marimont show how removing diseases for which a link between smoking and mortality has been alleged but not proven cuts the hypothetical number of smoking-related fatalities in half.
Replacing an unrealistically low death rate for never-smokers with the real fatality rate cuts the number by another third.
Controlling for “confounding factors”—such as the fact that smokers tend to exercise less, drink more, and accept high-risk jobs—reduces it by about half again. Instead of 400,000 smoking-related deaths a year, Levy and Marimont estimate the number to be around 100,000.
This would place the lifetime odds of dying from smoking at 6 to 1 (45 million smokers divided by 100,000 deaths per year x 75 years), rather than 3 to 1. However, about half (45 percent) of all smoking-related deaths occur at age 75 or higher. These can be called “premature” only by stretching common usage of the word. The odds of a life-long smoker dying prematurely of a smoking-related disease, then, are about 12 to 1.
A billboard that reads “Odds of dying in a car crash: 90 - 1; dying from smoking, 12 - 1,” would more accurately convey the truth than the current model, but might be less likely to capture our attention. Is it okay for health officials to lie to get our attention?