higher exposures to tar and specific carcinogens (e.g., tobacco-specific nitrosamines and benzo[a]pyrene) (Djordjevic et al., 2000; Fischer et al., 1989; Hoffman and Hoffman, 1997). For example, using modified protocols to stimulate human smoking behavior, the medium-yield (0.9–1.2 mg nicotine per cigarette) and low-yield (0.8 mg nicotine per cigarette) cigarettes deliver similar amounts of tar per day, although by FTC method measured per cigarette yields of tar, benzo[a]pyrene (BaP), and tobacco-specific nitrosamines (TSNAs) were higher in the former (Djordjevic et al., 2000). As cigarettes with different designs are developed and marketed, an assumption that the FTC method of estimating yields will be comparable to existing products is premature.

Over the last 30 years, data from surveys have been an important tool in the assessment of tobacco exposure among individuals and the population. They have been an effective means of tracking patterns of tobacco use and the societal perceptions that ultimately influence consumption. Individual exposure can be assessed through the measurement of the number of cigarettes smoked per day, duration of smoking, types or brands of cigarettes smoked (e.g., “tar” delivery, filter type, type of tobacco, mentholation), and age at initiation (IARC, 1986; Kaufman et al., 1989; La Vecchia et al., 1990; Lubin et al., 1984; Stellman and Garfinkel, 1989; U.S. DHHS, 1988; Vutuc and Kunze, 1983; Wilcox et al., 1988; Zang and Wynder, 1992). Lifetime exposures can be estimated by calculating pack-years (average packs per day multiplied by number of years smoked) or cumulative tar exposure (Zang and Wynder, 1992). A more detailed description of the most common surveys in use is presented in Table 11–5.

Most analyses indicate that self-report validity among adults is good (Patrick et al., 1994). Certain limitations, however, are evident in this type of exposure assessment (Giovino, 1999; U.S.DHHS, 1994). First, sampling errors may occur in any study in which generalizations are made from a selected population sample. One example is the over- or underrepresentation of certain groups, especially those that exhibit significant tobacco use or have differing smoking behavior. In fact, there is a built-in exclusion in many of the major surveillance tools of various segments of the population, such as the institutionalized mentally ill, prisoners, and those in areas of inadequate telephone coverage. Errors in response must be considered including memory errors, nonresponse errors, and misclassifications and inconsistencies in reporting. The validity of self-reported responses can be influenced by many factors (Velicer et al., 1992), particularly the respondent’s perception of privacy (Giovino, 1999). This is especially a concern among adolescents in the home setting and among groups that have increased pressure to abstain or to quit, including pregnant women, adolescents, and patients with heart or lung disease. One

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement