Tobacco Documents Online

3 established brands (Owen, 1976; Gori, 1980; U.S. Department of Health and Human Services, 1981). The less hazardous cigarette would seem to be the solution to a number of dilemmas. For the smoker who is both strongly attracted or habituated to continued smoking but also desires to reduce the hazard to his or her health, it offers the promise of compromise. For. the manufacturers, reducing T/N has proved an important marketing tool to-reach an increasingly health-conscious public and to reduce criticism in the biomedical community, without serious economic loss to interests dependent on tobacco sales. The less hazardous cigar- ette might be a compromise between the statutory commitments of the federal government to public health (and thus antismoking efforts) on one hand and to agriculture and-other economic activity on the other. The logic that lower T/N yields equal less harmful smoking seems simple and persuasive. But there are two' ways in which this logic may be misleading. First, the measurements of T/N are performed, in the laboratories of the Federal'Trade Commission (FTC) and others, by analyzing batches of smoke drawn by a machine that simulates smoking with a simple and unchanging program (Kozlowski et al., 1980; RKozlowski, 1981). Human smokers and their cigarettes, however, are neither simple nor unchanging. If lowering the T/N of cigarettes typically results in people's smoking more of them or smoking them differently, then the machine results may not predict the human results. Second, there is the complexity of the product itself. Tobacco smoke contains several thousand distinct compounds (Guerin, 1980). While the particu- late condensate we call tar is clearly carcinogenic, and pure nicotine has well-demonstrated effects on the cardiovascular system, the rated quantities of these two components (that is, the T/N yields given by the Federal Trade Commission and related methods) cannot give all the information relevant to the potential toxicity of cigarettes. In particular, these rating s do not take into account the yield of gases--such as carbon monoxide, hydrogen cyanide, and acrolein--in cigarette smoke, which may not parallel T/N yields as the cigarette is smoked. Ir. addition, flavorings are added to tobacco to modify the taste for consumer satisfaction. As a rule these additives are not under the purview of federal regulation and are held as TIMN 308074

4 industrial secrets. It is possible that some flavorings designed to offset reduced T/N taste may prove to be added risk factors. A variety of studies has been published regarding the toxicity of cigarette smoke, the different ways smokers puff on cigarettes, and the effects in human subjects of smoking lower•versus higher T/N cigarettes. Our report was undertaken largely to address the results of these studies, to try•to distill from them some up-to-date advice to smokers, to the government, and to researchers concerning the degree to which alternatives to smoking cessation--prinicipally the use of less hazardous cigarettes-are protective of health. The body of this report is an examination of relevant findings and theories and recommendations for a research agenda. From our review of data and concepts we have formu- lated some summary conclusions for smokers and for the government. • Despite the adverse consequences for some people of quitting cigarette smoking, such as weight gain and psychological distress, we are convinced by the evidence that habitual cigarette smoking is unequivocally hazardous to health and that longevity can be enhanced by stopping. Smokers who want to reduce the health hazards from their cigarettes are best advised to auit smoking entirely. • Short of cessation, reducing the intake of hazardous combustion products of cigarette smoking should reduce the health hazard. Switching to a lower tar and nicotine brand or cutting down the number of cigarettes e smoked of the same brand could achieve this reduction, but the effects of changing the brand or the number of cigarettes smoked are complicated. For continuing smokers, exposure to the constituents of smoke and attendant risks depend not only on the content, construction, and number of cigarettes but also on the way they are smoked. , • Smokers have not been educated about the meaning of the T/N information that appears on packages of most "light" and "low"- yield TIMN 308075

T/N brands as well as in advertising. These numbers derive from standard measurements on a calibrated smoking machine. The machine is set to smoke in a uniform way, whereas smokers exhibit many different patterns. There is no easy way to represent the variability of a smoking population when making such measure- ments. In addition to nicotine and tar, cigarette smoking delivers carbon monoxide and other toxic gases to the body, which are not currently measured. Therefore, the T/N yields may or may not coriespond well to the actual hazard exposure of different smokers. • There is evidence that many smokers respond to switching to lower-yield brands, at least in the short term, by altering their smoking patterns; they may increase their depth of inhalation, puff frequency, duration of holding smoke in the lungs, and/or number of cigarettes consumed. These char.ges tend to offset to some degree the expected reductions in nicotine and tar delivered by the new brands; the changes may also be difficult for the smoker to detect. Experimental evidence indicates that when such increases occur, smokers still may not absorb from the lower- yield brands all of the T/N previously absorbed from higher-yield brands. However, exposure to carbon monoxide and, by implica- tion, to other gaseous components generally seems to stay about the same. While some large-scale studies have suggested small gains in health due to using lower T/N (or filter rather than nonfilter) cigarettes, other population-wide studies do not support this view. Thus, the evidence for switching to lower T/N cigarettes is doubtful. • Many smokers may switch brands rather than quit smoking in the belief that their health gains will be essentially proportional to the reduction in rated TIN yield; or will be substantial; or that the lower-yield brand of their choice is virtually safe. In our ;udament, the degree of benefit most smokers can expect from switching to lower T/N brands, if any, is small compared with the benefit of stopvina smoking comoletelv. TIMN 308016

6 • Finally, the attempt to make cigarettes less hazardous by reducing their tar and nicotine yield is necessarily a crude approach, corres- ponding to the uncertainty of knowledge about the differential effects of the thousands of components of cigarette smoke and the difficulties in independently manipulating them. The weight of recent studies that we have reviewed supports the idea that nicotine dependence is a very important component of smoking behavior and that most heavy cigarette smokers, regardless of brand, tend to maintain high nicotine levels. For such smokers, studies should be conducted on the relative risks and benefits of tobacco products that deliver nicotine less encumbered by additional toxic compounds. The basis for these conclusions is an assessment of two related bodies of f indings, which are presented below: (1) epidemiological studies on the health consequences of cigarette smoking and.(2).laboratory and field studies of smoking behavior involving detailed analyses of the quantity, frequency, and mechanisms of tobacco smoke inhalation. The epidemiological studies strongly support a monotonic dose/response relationship between the number of cigarettes smoked and the emergence of serious cardiovascular, pulmonary, and other diseases (U.S. Department of Health, Education, and Welfare, 1979). However, the studies do not as a whole substantiate the equation of lower TIN with healthier smokers. The laboratory and field studies of smoking generally but not uniformly show that a significant proportion of smokers who switch to lower T/N cigarettes change the way they smoke them so as to compensate partially for the reductions in T/N yields, thus making the reduction in actual T/N absorption by smokers less substantial than the reduction in TIN ratings based on the FTC smoking machine method. These studies also indicate that yields of other possibly harmful components of inhaled smoke, such as carbon monoxide, are not reflected by T/N ratings. The review of these studies is followed by an assessment of current research needs on alternatives to smoking cessation. TIM~ 3og0'1'1

THE HEALTH EFFECTS OF CIGARETTE SMOKING EARLIER STUDIES Habitual cigarette smoking is hazardous to health. Morbidity and mortality data from epidemiological studies have shown convincingly that there is a strong relationship between smoking and increased death rates, particularly from•lung and other cancers and cardio- vascular disease. The health prognoses for smokers and nonsmokers are so different that a middle-aged, male, pack-and-a-half-a-day cigarette smoker who began smoking in his teens has a life expectancy roughly one decade less than a man matched on a variety of relevant characteristics who has never smoked. Ex-smokers also gain a mortality advantage over continuing smokers: it begins soon after quitting and increases with years of abstinence up to 20 years, when the mortality of ex-smokers is indistinguishable from lifetime nonsmokers (U.S. Department of Flealth, Education, and Welfare, 1979). It is more difficult to draw conclusions about the health differences between different sorts of smokers, beyond the well-confirmed proposition that smoking few cigarettes, other things being equal, is less harmful than smoking many. This has been demonstrated for the overall risk of death, the risk of death from cardio- vascular diseases, and for the risk of developing and dying from neoplastic diseases of the respiratory system (U.S. Department of Health, Education, and Welfare, 1979). ' In the largest prospective study, which initially enrolled 847,825 subjects, Hammond et al. report that, after statistically matching subgroups of smokers in their sample, "the adjusted number of (lung cancer] 7 TIMN 308078

8 deaths in low T/N smokers ranged from 81% to 88% of the adjusted number of deaths in high T/N smokers" (1977:107-108). The matching procedure was designed to hold the number of cigarettes smoked daily, among other variables, constant. Wynder and Stellman (1979), in a matched case/control study of patients with lung and larynx cancer, also reported statistically significant differences in favor of smokers of filter versus nonfilter cigarettes, with the number smoked daily held constant. Auerbach et al. (1979) examined sections of lung tissue from postmortem.examinations of more than 200 smokers who died between 1955 and 1959 (deaths other than lung cancer) and a comparison group who died between 1970 and 1977. The number of cell anomalies, some of which the authors hypothesize could be precancerous, was far lower among those smoking similar numbers of cigarettes in the 1970-1977 group than the 1955-1959 group. The authors attribute this to the trend toward lower-yield commercial cigarettes. A large prospective study in Scotland reported .filter cigarettes and lower T/N yields to be associated with lower prevalence of respiratory illness but not with decreased death rates (Hawthorne and Fry, 1978). Another prospective study in Great Britain (Higgenbottam et al., 1980) found only small differences in lung function across varying T/N yields, differences that nearly vanished among smokers of more than one pack daily. A third British study (Wald, 1976) reported lower T/N levels to be correlated with reductions in pulmonary but not cardiovascular diseases. A recent report from the Framingham study (Castelli et al., 1981) found that smokers of nonfilter cigarettes had slightly lower coronary heart disease morbidity than smokers of filter cigarettes, which are generally lower in T/N yield. Lee and Garfinkel reviewed these (except for Castelli et al.) and other epidemiological studies and concluded nevertheless that "smokers of filter (or low T/N) cigarettes have lower mortality than smokers of plain (or high T/N) cigarettes for those diseases most strongly associated with smoking . . ." (Lee and Garfinkel, 1980:23). This result occurs principally in studies in which the number of cigarettes smoked daily is statistically controlled, i.e., when smokers of 10-19 filter cigarettes daily are compared only with smokers of 10-19 nonfilter cigarettes; smokers of 20-29 TIMN 308079

9 filter cigarettes with smokers of 20-29 nonfilters; and so forth. Lee and Garfinkel note that this result would be somewhat misleading if in fact smokers of low T/N or f ilter cigarettes tended as a rule to smoke more cigar- ettes daily than smokers of high T/N or nonfilter cigarettes. To counter this difficulty, they cite Garfinkel's analysis of data that led him to conclude that "over a long period of time, people tend to smoke the same number of cigarettes a day regardless of tar and nicotine level" (Garfinkel, 1980:24). However, in this analysis, only one-third of continuing smokers said that they smoked the same number after 13 years, and small but consistent differences were seen in changes in the number of cigarettes smoked across time depending on T/N, even when the changes in T/N and the number of cigarettes smoked were collapsed into broad categories. The overall consumption of T/N per capita among U.S. adults, based on the total numbers of cigarettes sold, their FTC machine-measured T/N ratings, and census population figures, is calculated to have declined by about half between 1955 and 1975, and most of this ' reduction occurred by 1966 (Wakeham, 1976). It seems to us that this reduction provides a natural experiment for evaluating the health effects of reduced-yield cigarettes on the adult U.S. population. Because of the long period needed for the development of morbidity and mortality differences, it may be too early to determine whether these changes in T/N yields were favorable for those who have smoked cigarettes only since 1965, i.e., principally smokers born after 1945. In addition, since 1975 there has been a notable increase in the sales of "ultra-low" T/N brands, i.e., those yielding less than 5 mg. of tar and 0.5 of nicotine. However, we should be able to detect the impact of the newer cigarette;products through 1975 on smokers born before 1945, by analyzing appropriate trends in U.S. health statistics relative to available data on smoking patterns. The largest share of morbidity and mortality attributable to smoking is due to its elevation of the risk of cardiovascular illness, including athero- sclerosis, myocardial infarction, coronary heart disease, sudden cardiac death, and peripheral vascular disease (U.S. Department of Health, Education, and Welfare, 1979). Smoking is one of several major cardio- TIMN 308080

10 vascular risk factors, which include such behavioral patterns as exercise, stress, and diet. The plurality of attributable risk factors complicates the detection of differences in risk among smokers of cigarettes with differing T/N. Moreover, the chances of dying from many of these disease have been reduced in recent years by declines in the incidence of several of the risk factors as well as better medical care. A more easily analyzed health effect from the point of view of evaluating reduced T/N yield is cancer of the lungs and bronchi. There is no widespread behavioral risk factor for these cancers that is comparable, in either relative or absolute effect, to smoking cigar- ettes. Although the total burden of illness due to cardiovascular disease is many times larger, the loss due to these cancers is considerable--more than 100,000 .new cases annually in the United States. Most of these cancers are attributable to cigarette smoking and most still lie beyond the reach of lifesaving medical intervention. The following analysis, undertaken for this report, addresses the effect of reduced T/N yields on respiratory system cancer. CHANGES IN DEATH RATES FROM RESPIRATORY SYSTEM CANCER IN THE UNITED STATES SINCE 1940 Among men between ages 25 and 65 in a given year, there has been a notable decline since 1955 in the proportion who are currently smoking. As of 1975 (the latest year that complete data for this analysis were available), there were about a fourth fewer cigarette smokers in each five-year age bracket than had been the case for men at the same ages 20 years before (see Figure 1). However; the proportion of the overall male population who were current heavy smokers (defined as more than a pack of cigarettes daily) did not decline during these years. Most of the drop-off has been due to the replacement of light smokers in the male population by nonsmokers. The proportion of current smokers who smoke heavily has therefore increased, while by 1975 nonsmokers had become the majority of men at every age. The risk of death from lung cancer is 10 to 20 times greater among current heavy smokers than for their nonsmoking age-mates, accounting for between TIMN 308081

11 Smok.rf of mort than ont pack daily AGE FIGURE 1 Estimated Prevalence of Current Cigarette Smoking by U.S. Men, 1955, 1965, 1975 Sources: Haenszel et al. (1956); Ahmed and Gleason (1970); USDHEW (1976, 1980). one-half and two-thirds of all male lung cancer deaths (lighter smokers account for most of the remaining lung cancer-deaths). Since heavy smokers have maintained their overall proportion of about one in six in the male population ages 25-65 and since these smokers account for the bulk of lung cancer deaths, we can use the annual age-specific death rates to see whether a discernible improvement in lung cancer mortality has occurred, as might be anticipated with the lower T/N cigarettes that have become available since 1955. It should be kept in mind that this is the one category of health effect for which the 1981 report of the Surgeon TIMN 308082

12 General held open the possibility that lower T/N cigarettes might be relatively less hazardous (U.S. Department of Health and Human Sirvices, 1981). We have analyzed a data series on deaths due to respiratory system cancers (RSC), categories 160-165 in the ninth revision of the International Classification of Diseases. Most of the deaths in the data series are due to lung cancer (see Wynder et al., 1970; Enstrom and Godley, 1980). A small fraction of RSC deaths cannot be attributed to the effects of cigarette smokingt these data are, however, the best general indicator of changes in smoking-related health damage that is available across the periods of time necessary to this inquiry. We return to the difficulties of interpreting these data below. Despite the stability in the proportions of the adult male population who are current heavy smokers in each age category and despite the substantial reductions in the proportions of lighter smokers, there have been large increases in death rates from respiratory system ;ancer since 1955 among men ages 35-65, and the rates among younger men-begin to level off or fall slightly only after 1970 (see Figure 2). The overall increase has been on the order of 70 percent more RSC deaths per capita (males, age-specific) in 1975 than in 1955. These increases are at least partially attributable to two complicating factors: the steady rise in the prevalence of heavy smoking among these men prior to 1955 (Horn, 1977) and the fact that development of the more serious health effects of smoking generally requires periods of time. The conjunction of these factors makes the use of current or recent smoking patterns insufficient to sort out the RSC effects of cigarettes (Burch, 1980). Based on surveys from the National Clearinghouse for Smoking and Health and others, Horn (1977) has published estimates of accumulated c;garettes smoked per capita in terms of total lifetime cigarette packs for virtually the same male age cohorts. Using these figures it is possible to construct an inc',ex of the relationship between lifetime cigarettes smoked and RSC deaths. The results of this calculation, reported in Table 1, can be thought of as roughly tracking the carcinogenicity of cigarettes as smoked by American men, controlling for the aggregate, lifetime numbers of TIMN 308083