Smoking and Tobacco Control Monograph No. 7 
among men who have high blood pressure or diabetes and appears to be 
a consequence of increased atherosclerotic disease in the blood vessels 
supplying the genitalia rather than an effect on sexual drive. 
ABSOLUTE RISK Human epidemiology can be used to estimate quantitatively the 
VS. RELATIVE risk of specific diseases to human smokers. For example, in the 
RISK CPS-11 study of smoking practices and mortality rates among 
1.2 million U.S. adults followed from 1982 through 1986, about 0.8 percent 
of current male smokers ages 65 or older died of lung cancer each year (U.S. 
Department of Health and Human Services, 1989), whereas the comparable 
annual lung cancer death rate was about 0.04 percent among men ages 65 or 
older who never smoked. These quantitative risk estimates are often termed 
"absolute risks." That the continuing smokers' risk of lung cancer was 
twentyfold that of nonsmokers is an expression of "relative risk." 
Estimating relative risks from analyses of chemical composition of 
different cigarettes is far more complicated. For example, the smoke from 
cigarette A might contain 0.05 mg of BaP, a known carcinogen, whereas 
the smoke from cigarette B might contain 0.02 mg of BaP. To estimate 
human lung cancer risks from these data alone would require a number of 
assumptions relating the dose of BaP to the incidence lung cancer in humans. 
Whereas cigarette A had 2.5-fold as much BaP as cigarette B, it cannot be 
concluded automatically that the relative risk of getting lung cancer for 
those smoking cigarette A is 2.5-fold greater than those smoking cigarette B. 
The relative concentrations of benz(a)anthracene, another carcinogen in the 
PAH group, might be higher or lower. 
Toxicity studies in nonhuman species also can give estimates of 
relative risk, but applying these estimates directly to humans requires 
caution. The fact that the smoke from cigarette C might produce twice as 
many revertants as cigarette D in a particular strain of the Ames salmonella 
assay is an indicator that C contains higher concentrations of certain 
mutagens. Likewise, if cigarette E produced three times as many tumors as 
cigarette F in a mouse skin carcinogenesis assay, we can conclude that 
cigarette E contains higher concentrations of certain carcinogens, including 
tumor initiators and tumor promotors (DuMouchel and Harris, 1983). 
TAR, NICOTINE, Some studies (e.g., Adams et al., 1987) suggest that the yields 
CARBON MONOXIDE, of most toxic agents in cigarette smoke are correlated with 
AND OTHER SMOKE their tar, nicotine, and CO deliveries. Still other studies 
CONSTITUENTS show the correlation to be weak at best. Kaiserman and 
Rickert (1992) found a 0.89 correlation between the declared tar level and 
the BaP delivery of 35 brands of Canadian cigarettes. However, for 16-mg 
tar brands, the measured BaP ranged from 15 to 28 ng per cigarette. Fischer 
and colleagues (1991) found no correlation between tar delivery and the 
concentration of certain TSNAs in 170 Imropean cigarettes. 
I he lack of a perfect correlation between tar values and specific chemical 
yields is not simply an artifact of measurement error. As Hoffmann and 
colleagues (this volume) report, there are many alternative methods to reduce 
66 
