Smoking and Tobacco Control Monograph No. 7 
health relevance: whether disease risk varies with cigarette tar and nicotine 
yield as determined by the FTC method. 
Exposure misclassification is a potential threat to the validity of studies I 
of the changing cigarette. Typically, the exposure of smokers to cigarettes I 
of varying tar and nicotine yields is estimated based on information on | 
brands and types of cigarettes smoked. However, smokers may not be able ; 
to provide a fully accurate history of brands used throughout their lifetimes; , 
therefore, estimates of tar and nicotine yield are potentially subject to error 
(U.S. Department of Health and Human Services, 1990). The consequences 
of misclassification include biased estimates of the effect of tar and nicotine 
yields and reduction of statistical power. Additional methodological concerns I 
include the possibility of selection bias if smokers affected by symptoms 
or disease tend to switch to lower yield products; another concern is 
confounding by other aspects of lifestyle if smokers of lower yield cigarettes j 
differ substantially in lifestyle characteristics from those smoking higher yield ; 
products. However, the research challenge of studying the consequences of 
the changing cigarette is no different from the challenge posed by other j 
complex mixtures of inhaled agents, and epidemiologic research has the j 
advantage of integrating the effects of the mixture, even though individual j 
components may be interacting in ways that are difficult to characterize. | 
I 
Lung Cancer The ACS's Cancer Prevention Study I (CPS-I) provided early evidence j 
on the risks of lower tar and nicotine cigarettes (Hammond et al., 1976). j 
CPS-1 included about 1 million volunteers who were followed from 1960 to i 
1972. Mortality was examined by three categories of tar intake — high, jj 
medium, and low. For all causes of mortality and for lung cancer mortality, jj. 
the standardized mortality ratios declined as estimated tar or nicotine intake jiii 
declined (Table 1). The findings were similar for males (Table 1) and for jl 
females (data not shown). However, comparison with mortality in never- i| 
smokers shows that smokers of even the lowest tar and nicotine products ‘j 
nonetheless had substantially higher mortality rates. : 
Other studies have had similar findings for lung cancer. Wynder and jf 
colleagues at the American Health Foundation have conducted an ongoing ! 
case-control study of smoking and lung cancer that provides information on ' 
cigarette type and lung cancer risk over decades since the 1950's. Reports i 
from this study have consistently shown that smokers of lower tar products, I 
indexed in a variety of ways, have reduced lung cancer risk (Wynder et al., : 
1970; Wynder and Rabat, 1988). For example, in a recent report based on 
cases from the late 1970's and early 1980's, risks were examined separately ■; ! 
for persons with squamous cell and sniall cell carcinomas of the lung j' 
(Kreyberg 1) and adenocarcinoma of the lung (Kreyberg 11) (Wynder and 
Rabat, 1988). .Smoking was classified as 100 percent filter, 100 percent j 
nonfilter, or intermediate, by number of switchers from nonfilter to filter. 
For smokers of filter cigarettes only, risks were approximately 10 to 30 percent 
less than those of smokers of nonfilters only ( Fable 2). I 
I' 
