Tobacco Documents Online

The value of preventive medicine _'stablished in IBA-GEIGY medical and each year on papers and ,n also holds ide scientific Ciba Foundation Symposium 110 ~11 the rides on in the house UNIVERSITY of NORTH CAROLINA 1985 Pitman London NAR 14 1985 HEALTH SCIENCES LIBRARY

Control of tobacco-related disease RICHARD PETO Cancer Studies Unit, Radcliffe Infirmary, Oxford OX2 6HE, UK Abstract. As ways of discouraging tobacco consumption, the effects of increases in price and in quantitative information may have been under-emphasized. To decrease the hazards of tobacco, switches from cigarettes to pipes, cigars or 'smokeless' tobacco may be useful, as may a reduction in cigarette tar delivery. Indeed, the spread of existing tar level reductions from capitalist to socialist countries might prevent tens of thousands of lung cancer deaths each year in the early decades of the next century, and (perhaps by attempts to engineer cigarettes so that smokers of lower tar cigarettes are less likely to "compensate' by taking more smoke) it should eventually also be possible to change cigarettes so as also to reduce their effects on heart and lung disease. Changes in consumption and in composition of tobacco products are complementary, not competing, strategies. If both are pursued effec- tively, then although the life expectancy of old people may not be much improved, the proportion of adults who die before reaching old age will decrease substantially. 1985 The value of preventive medicine. Pitman, London ( Ciba Foundation symposium 110) p 126-142 The obvious way to avoid tobacco-related disease is to avoid smoking, and eventually the most important way of controlling tobacco-related disease will be by vast reductions in the extent to which tobacco, particularly in the form of cigarettes, is smoked. But at the moment, although cigarette consumption is going down in certain countries, worldwide cigarette consumption is going up. It is therefore an extremely destructive form of idealism to consider only the discouragement of cigarette consumption but to avoid considering other ways of modifying the extent to which cigarettes kill people. There are, at least in principle, three different ways of reducing tobacco-related disease. First, one may be able to modify the "host'--the smoker. Mention has been made at this symposium of the Japanese, who appear as a nation to eat less fat than we in Britain, and who moreover appear to eat fat with a higher "P:S ratio' (i.e. with a higher ratio of polyunsaturated to saturated fats). Although smoking has been widely prevalent in Japan for more than 30 years, Japanese death rates from heart disease are still, even in early middle age, very much lower than those in the US or Britain. In other words, the absolute heart disease risk per cigarette is much lower in Japan than in Britain or America. 126 TOBACCO-Rt Studies of Ja: non-genetic A second particular eit some less ha: changes in c filter tips envisaged. British gove: Health (1982- cigarette, bt about the c~ tobacco in a alternatives tobacco (inc is, powderec exact hazart liminary evit unlikely to produced b~ the young er The third sumption. Ir because wid risks are un. two hundred continued to moment, ho must also cc tobacco prot on other top consumptior Changes in tt Probably in, aged: from c haps. cigars) consider the merely becat

ase lects of increases in price and . To decreabe the hazards of ,s tobacco may be useful, as existing tar level reductions asands of tung cancer deaths ups by attempts to engineer x to "compensate" by taking garette550 as also to reduce ,tion and in composition of :s. If both are pursued effec- not be much improved, the •ase substantially. rton symposium 110) O avoid smoking, and • co-related disease will • ticularly in the form of zarette consumption is nsumption is going up. rn to consider only the :onsidering other ways • There are, at least in ed disease. ker. Mention has been • a nation to eat less fat vith a higher 'P:S ratio' rated fats). Although aan 30 years, Japanese niddle age, very much :Is, the absolute heart n Britain or America. TOBACCO-RELATED DISEASE 127 Studies of Japanese in America suggest that these differences are due chiefly to non-genetic 'host factors'. A second way is to affect the manner in which tobacco is used, and in particular either to encourage a switch from cigarettes to the use of tobacco in some less hazardous form, or to modify the composition of cigarettes. The chief changes in cigarette composition thus far have involved the introduction of filter tips and the lowering of tar levels, but other modifications could be envisaged. One possible change was discussed in the most recent report to the British government of the Independent Scientific Committee on Tobacco and Health (1983). This was the possible introduction not of a low-nicotine, low-tar cigarette, but of a medium-nicotine, low-tar cigarette. As well as thinking about the composition of cigarettes, we should also think about the use of tobacco in a much less hazardous form, including perhaps not only the usual alternatives such as pipes and cigars, but also various forms of 'smokeless' tobacco (including nasal snuff, chewing tobacco, and 'dipping' tobacco--that is, powdered tobacco that is usually held between the gum and cheek). The exact hazards associated with such habits are not yet known, but the pre- liminary evidence thus far available (Winnet al 1981) suggests that they are unlikely to be substantial in comparison with the vast mortality now being produced by tobacco smoking (unless promotion of smokeless tobacco among the young engenders nicotine addiction that eventually leads to smoking). The third and most important way is, of course, to decrease tobacco cor;- sumption. In the end, this is going to be the solution that is finally adopted, because widespread tobacco use is (at least if the product is smoked and the risks are understood) unacceptably hazardous. Eventually--perhaps one or two hundred years from now--people will perhaps find it amazing that tobacco continued to be smoked so widely in the second half of the 20th century. For the moment, however, cigarette consumption is going up Worldwide, and so one must also consider the first two factors (modification of the host, and of the tobacco product being used). I will leave modification of the host to speakers on other topics, however, and will be concerned only with the composition and consumption of tobacco products. Changes in the way in which tobacco is used Probably in decreasing order of efficacy, three types of change may be envis- aged: from cigarettes to smokeless tobacco, from cigarettes to pipes (or, per- haps, cigars), and from more hazardous to less hazardous cigarettes. I shall consider the latter at greatest length, not because it is the most effective, but merely because politically it may be the easiest to achieve.

128 PETO TOB~~ Smokeless tobacco In the South-Eastern United States, many women have, throughout their adult lives, habitually 'dipped snuff--that is to say, they have placed powdered tobacco between the gum and cheek, and have thereby absorbed a number of pharmacologically active substances. It is noteworthy that among snuff- dippers in the South-Eastern US. the proportion who smoke (15%) is much lower than that among other women (45%) (Winn et al 1981). Moreover, even among those who do smoke the consumption of cigarettes per smoker is slightly lower among the "dippers'. These observations suggest that snuff dipping discourages smoking in that particular population, giving some hope that it might also do so elsewhere. Vigorous commercial promotion of snuff dipping has begun in America, and is just beginning in Europe. If this or some other such habit were to become widespread and did to any substantial extent replace smoking (particularly of cigarettes), then the net effect would be likely to be a reduction in tobacco-induced mortality. For, although snuff dipping causes a vast increase in the relative risk of cancer of the gum and cheek (together with the same sort of risks of cancers of other parts of the mouth that smoking produces), the absolute excess risks of death from oral cancer associated with the habit in the South-Eastern United States appear to be at most a few per cent of the total risk of death produced by cigarette smoking (Winn et al 1981). Although the absolute risks in other populations might, of course, be con- siderably different (especially if some diseases other than oral cancer are found to be increased by tobacco 'dipping'), the use of smokeless tobacco is still likely to be much less hazardous than is tobacco smoking, especially of cigarettes. Switch of smoking from cigarettes to pipes and/or cigars In prospective observations of male British doctors, (i) lifelong non-smokers, (ii) men who currently smoked only pipes and/or cigars (most of whom smoked only pipes), and (iii) men who currently smoked cigarettes had age- standardized death rates in the ratio 1: 1.09:1.64 (Doll & Peto 1976). This may not exactly reflect the relative hazards of the two forms of tobacco, but, to- gether with much other evidence, it does correctly suggest that habitual use of pipes is much less hazardous than is habitual use of cigarettes. Since, more- over, about half of the men who smoked only pipes and/or cigars had pre- viously smoked at least some cigarettes, it also strongly suggests that a wide- spread switch from cigarettes to pipes could save many lives, even though men who switch may inhale their pipe smoke in ways that are more hazardous than the usual inhalation patterns among lifelong pipe smokers. Chan Sever prod1 takin This effec~ smok lung than ~ less h Thirc cigar Th cflnc~ evidt the cigar (i) L Lun smo adul of c: age age redti rele 193~ (i.e Scar 195t per the por, to e unl~ corf

PETO hroughout their adult rye placed powdered • absorbed a number iv that among snuff- moke (1597) is much ~81). Moreover, even per smoker is slightly ,t that snuff dipping ng some hope that it orion of snuff dipping If this or some other atantial extent replace ould be likely to be a auff dipping causes a cheek (together with mouth that smoking incer associated with at most a few per cent g (Winn et al 1981). . li~ourse, be con- 3~lncer are found ; to5-acco is still likely ially of cigarettes. .-'elong non-smokers, ars (most of whom l cigarettes had age- )eto 1976). This may of tobacco, but, to- that habitual use of rettes. Since, more- l/or cigars had pre- uggests that a wide- Is, even though men rare hazardous than TOBACCO-RELATED DISEASE 129 Changing the composition of cigarettes Several questions are relevant. First, scientifically, how can a cigarette be produced that is less hazardous in use? (N.B. Smokers tend to 'compensate' by taking more smoke from today's low-tar than from the old high-tar cigarettes.) This first question may conveniently be divided into (a) assessment of the effects of changes that have already been introdiaced on the three principal smoking-related diseases (lung cancer, heart disease, and chronic obstructive lung disease), and (b) design of cigarettes that are likely to be less hazardous than those currently being sold. Second, politically, how can the switch towards less hazardous cigarettes be encouraged in Western and in Eastern countries? Third, psychologically, to what extent (if any) will the pursuit of less hazardous cigarettes impede measures to reduce cigarette sales? The three main diseases associated with cigarette smoking are (i) lung cancer, (ii) heart disease, and (iii) chronic obstructive lung disease. For lung cancer, but not for the other two diseases, there is now reasonably good evidence that the changes in cigarette composition that have taken place over the last few decades in Western countries have reduced the hazard per cigarette. (i) Lung cancer Lung cancer risks in adult life depend surprisingly strongly not only on recent smoking habits, but also on smoking habits many decades beforehand, in early adult life. Thus, for example, among 60-year-old habitual smokers of one pack of cigarettes per day, those who began to smoke cigarettes regularly at about age 15 have more than twice the lung cancer risks of those who began at about age 25 (Doll & Peto 1981, Appendix E). This suggests that if the tar level reductions have any worthwhile effect, then tar levels in early adult life may be relevant to lung cancer risks in middle age, mar.y decades later. Between the 1930s and the 1970s there have been reductions of more than 50% in the mean (i.e. sales-weighted) tar delivery per cigarette in the United States, Britain, Scandinavia, and a few other places. These changes were small until the late 1950s and then they suddenly became rapid, with decreases from 30-odd mg per cigarette in the mid-1950s down to approximately 15 mg per cigarette by the 1970s. The changes are not expensive and involve the use of filter-tips, porous paper (or even. as an extreme measure, 'ventilated' filters that allow air to enter into the side of the filter to dilute the smoke) and modified types of tobacco (which may in some instances actually be less .expensive than unmodified tobacco). There is, of course, a reduction not only in the unwanted components of the smoke but also in those substances (e.g. nicotine?) to which

130 PETO TOBACCO-RE[ some smokers are addicted, and when such reductions occur many smokers are likely to compensate, either by smoking more cigarettes* or, perhaps more commonly, by taking in more smoke per cigaretteS-. It appears, however, that the latter form of compensation is not always sufficient to outweigh the reduc- tion in tar (Wald et a11980), in which case the net result will be inhalation of less tar into the lung. This conclusion is suggested both by common sense and by observation, but even if it is accepted it does not prove that the hazards will be correspondingly reduced, for despite some 30 years of laboratory research the importantly carcinogenic factors in cigarette smoke have not yet been identified reliably. Moreover, it is difficult to predict how changed patterns of inhalation will change what is deposited on the main target areas--which, for lung cancer, are not the peripheral tissues, but the large airways--as the smoke streams past them. Consequently. it is necessary to discover by direct epidem- iological observation whether the risks of lung cancer are materially reduced by the widespread switch to lower tar cigarettes. Unfortunately, this is not easy to do, for not only are smokers of low-tar brands self-selected but also, just as it is only after some decades of smoking that the full risks materialize, so perhaps it is only after some decades of using low-tar cigarettes that the full risk avoidance will materialize. Therefore, even if the effects in late middle age will one day be substantial, they may not yet be. Any substantial effects that are going to materialize in early middle age should be beginning to be evident by now in Britain, however, for although the tar reductions of the 1950s were only moderate, those of the 1960s were substantial in Britain, North America and Scandinavia. Thus, a 40-year-old in 1980 will have been smoking from about 1960 to 1980, throughout most of which time tar levels were substantially lower than in previous decades. Two main pieces of epidemiological evidence are currently available, the first being the results from classical case--control or prospective surveys. Unfor- * In principle, tar reductions could either increase or decrease the number of people who smoke (by making it less of an ordeal for non-smokers to acquire the habit or by making the habit less addictive) and could either increase or decrease the number of cigarettes one individual smoker consumes (by increasing the number needed to achieve a given dose or by decreasing the satisfac- tion per cigarette). In practice, however, the patterns of cigarette consumption in different coun- tries do not appear to be influenced consistently in either direction by changes in cigarette composition. "1" Surprisingly, there appears to be little reliable information on which of the many characteristics of the cigarette (e.g. nicotine, draw resistance, taste) importantly affect 'compensation'. If these could be identified and modified (e.g. by increasing the nicotine delivery, draw resistance or whatever of low-tar cigarettes), then maybe the intake of many toxins could be decreased simul- taneously. Such compensation presumably underlies the recent disappointing finding by Kaufman et al (1983) that the risks of myocardial infarction are not materially different among smokers of different types of cigarette. tunately, such relate chiefly t and even rece cigarettes for~ bated in studi~ the tar reducti. less extreme, t overall tar lev~ concurrent co with people o brands. Despi all the case-c~ that: 'a reasonabl tar/nicotine tar/nicotine smoking.. smoked the smoking liv~ even greate: improve ev~ Because of, concurrently, early middle a~ tes for much o can be supplen the study of na (Doll & Peto those from tht reductions are from the delay better to use t! began), Britist~ mately stabiliz and the reduct from changes i likely to be far in early middle Moreover, bot accelerating dc the next decad

PETO TOBACCO-RELATED DISEASE 131 , smokers are ,erhaps more ~owever, that ~h the reduc- ~lation of less .,ense and by tzards will be research the ot yet been ~d patterns of s--which, for -as the smoke :.irect epidem- 'ly reduced by is not easy to ,o. just as it is so perhaps it isk avoidance ill one day be are going to .nt by now in re only and g from about antially lower tvailable, the rveys. Unfor- ople who smoke ng the habit less dividual smoker sing the satisfac- ~ different coun- ges in cigarette ty characteristics ~sation'. If these aw resistance or :lecreased simul- ling by Kaufman nong smokers of tunately, such data as are currently available are limited by the fact that they relate chiefly to late middle or old age. when most of the lung cancers occur, and even recent studies relate chiefly to people who have smoked low-tar cigarettes for only a fraction of their smoking lives. This difficulty is exacer- bated in studies performed during the 1960s (or early 1970s) by the fact that the tar reductions then available for study were not only more recent, but also less extreme, than those now available. A related source bf difficulty is that as overall tar levels decrease, the higher tar levels simply cease to exist, so direct concurrent comparison of people now on low-tar cigarettes can be only with people on moderate-tar cigarettes, and not on the old very high tar brands. Despite these difficulties, when Lee & Garfinkel (198I) reviewed all the case-control and prospective studies then available they concluded that: 'a reasonably clear picture has emerged. This is that smokers of filter (or low tar/nicotine) cigarettes have a lower mortality than smokers of plain (or high tar/nicotine) cigarettes for those diseases most strongly associated with smoking... These reductions in mortality have been seen in those who have smoked the more modern types of cigarettes for only a small part of their smoking lives. The fact that those who have smoked them for longer show even greater reductions in mortality suggests that the overall picture will improve even more in years to come." Because of difficulties of self-selection, of comparing the new with the old concurrently, and of characterizing individuals' recent lung cancer rates in early middle age (i.e. the rates among people who have smoked low-tar cigaret- tes for much of their adult lives), the case-control and prospective survey data can be supplemented usefully by a second type of epidemiological data, namely the study of national trends in early middle age. However, for obvious reasons (Doll & Peto 1981). it is not advisable to use for this purpose data (such as those from the United States) in which any downward trends caused by tar reductions are likely to be diluted or even reversed by upward trends resulting from the delayed effects of past increases in tobacco consumption. Instead, it is better to use the British data. For, by the 1950s (when the rapid tar decreases began), British male lung cancer rates in early middle age had already approxi- mately stabilized (Table 1). Table 1 also describes their subsequent evolution, and the reductions are extremely impressive. They are most unlikely to result from changes in air pollution, for not only are any effects of the air pollution likely to be far smaller than this (Cederlrf et al 1978), but also similar halvings in early middle age have been seen over the last 20 years in unpolluted Finland. Moreover, both in Finland and in Britain the changes appear, if an.vthin~, to be accelerating downward, so if this pattern carries on into late middle age during the next decade or two, then at least in these two countries (where the male

132 PETO TABLE 1 Recent trends in England and Wales in male lung cancer death certification rates in early middle age"'b Death certification rates per million men from cancers of the respirator),' tract, excluding larynx Age 1951-1955 1956-1960 1983 Ratio (yrs) la) (b) (c) (c/b) 30-34 38c 37c 10 0.3 35-39 101¢ 95c 37 0.4 40-44 253c ~6~ 112 0.4 45-49 589¢ 597~ 2950 0.5 • Note both the approximate constancy before tar deliveries began to be greatly reduced and the large decrease thereafter. b Note: (1) These trends are not materially affected by changes in curative treatment of the disease. (2) Sales-weighted tar levels started to fail rapidly only at the end of the 1950s but are now less than half of what they then were. (3) Mean daily cigarette consumption per British male aged 30-50 did not change greatly until the past few years, and in 1955, 1965 and 1975 was respectively 10.5, 9.9 and 10.2 (Lee 1976): the 10-20% decrease in consumption that existed in the second half of the 1970s is too small and too late to be chiefly responsible for the large decreases in lung cancer mortality rates in 1983. (4) In unpolluted Finland, where male cigarette smoking also began so long ago that the trends had nearly flattened out by the late 1950s, male lung cancer rates in early middle age have likewise been approximately halved over the past 20 years, and in both countries the decreases appear, if anything, to be accelerating. (In Finland, as in Britain, no large changes in cigarette consumption were evident before the mid-1970s.) ¢ High mean tar intake throughout smoking history. d High intake only in first decade or so of smoking history. death rates are at present uniquely high) lung cancer may some day decrease for a few years* as fast as it once increased. (ii) Heart disease Perhaps because of the substantial extent to which smokers 'compensate' for tar delivery reductions, there is disappointingly little evidence of any favour- able effect of such reductions on heart disease. The studies reviewed by Lee & Garfinkel (1981) did in aggregate suggest some slight benefit, but the rather * It will not decrease to anywhere near non-smoker rates, however, unless there is widespread abandonment of cigarette smoking. Similarly, in those other populations where lung cancer rates have not yet completed their rise, even a tar-level reduction that halves the carcinogenicity of cigarettes may merely slow, rather than reverse, the progressive increase of the disease over the next few decades. TOBACCO better stuc no evident cancer ris numbers o rarity of ti slight but v disease (ai chief need tunately, t vestigated. If so, a I hazardous responsiblt ants, dra~ importantl minimize t This 'bla tant carci~ disease firs by random modified t\ The effe~ smoker is ~ she will ta difference study phar~ twice as mt There max only a fe~ nicotine, d, cigarette t~ tive lung di (iii) Chron, COLD is a ponse relat disease ma~ rates are ri that if we d should be a

PETO ifieation rates in early Rtltlo tc b~ o.3 0.4 (}.4 0.5 :atb reduced and the of the di.~ease. but are nov, less than ~nge greatly until the 10.2 {Lee 1976): the , is too small and too tv rates in 1983. :o that the trends had .,e have likewise been decreases appear, if garette consumption ne day decrease :ompensate' for .~ of any favour- iewed by Lee & • but the rather there is widespread re lung cancer rates : carcinogenicity of he disease over the TOBACCO-RELATED DISEASE 133 better stud)' of Kaufman et al (1983) suggests none. At least, however, there is no evidence of any adverse effect to set against the apparent reductions in lung cancer risk. and perhaps larger case-control studies, with even greater numbers of people in early middle age (i.e. in their forties or even, despite the rarity of the disease among young adults, in their thirties) will reveal some slight but worthwhile differences between one cigarette and another. For heart disease (and, probably, for chronic obstructive lung disease), however, the chief need is to design a cigarette that will minimize 'compensation'. Unfor- tunately, the exact determinants of compensation have not been properly in- vestigated. Perhaps. for example, many smokers smoke largely for nicotine. If so, a medium-nicotine, low-tar cigarette might be substantially less hazardous (for there is some evidence that nicotine itself is not chiefly responsible for the cardiotoxicity of cigarettes). Alternatively, perhaps flavor- ants, draw resistance, acidity variations or other manipulable factors might importantly affect "compensation', and might therefore be modifiable to minimize the amount of smoke taken per cigarette. This "black box" approach to cigarette design does not require that the impor- tant carcinogens, cardiotoxic agents and causes of chronic obstructive lung disease first be identified, and hence it could be pursued immediately, perhaps by randomized trials measuring compensation in smokers of various suitably modified types of cigarette. The effects of changes in tar delivery on compensation are quite marked: if a smoker is given a high-to-medium tar cigarette and a low-tar cigarette, he or she will take twice as much smoke from the low-tar cigarette. A twofold difference like that• which occurs immediately, should be relatively easy to study pharmacologically. What ingredient in the smoke tells the smoker to take twice as much from it? Is it the nicotine? Is it the draw resistance? Is it the taste? There may be several determinants of compensation, but there are probably only a few major ones. If we could identify one or two of them that, like nicotine, don't appear to be important toxins, it should be possible to produce a cigarette that would give less risk of lung cancer and less risk of chronic obstruc- tive lung disease. (iii) Chronic obstructive lung disease (COLD) COLD is as specifically related to smoking as is lung cancer, as the dose-res- ponse relationship among British doctors shows (Doll & Peto 1976). This disease may be decreasing in Britain, but in some countries, such as the US, the rates are rising rapidly. The dose-response relationship for COLD suggests that if we decrease the extent to which people take smoke from cigarettes, we should be able to produce a cigarette conferring less hazard of this disease. It is

134 PETO TOBACCO-REL. difficult to test this directly, however, because of the natural history of COLD (Peto et al 1983). We cannot use the approach used on trends for lung cancer, because although almost all cases are caused by tobacco, there are other impor- tant determinants of chronic obstructive lung disease. At a time when there were few marked differences in the smoking habits of the different social classes in the UK, COLD was much more common as a certified cau_se of death in the lower than in the upper social classes, in both men and women. Since then, from the late 1960s to the late 1970s, death rates among middle-aged men have halved from this disease in Britain, and are continuing to fall. It is very difficult to produce a specific explanation for these trends. One cannot confidently ascribe them to changes in cigarette composition. Air pollution decreases pro- vide an obvious explanation, but I am not sure that it is the correct one. Thus, the study of national mortality trends may not be directly informative about the effects of the post-1960 changes in cigarette composition on COLD mortality. Classical case-control studies are likely to be even less informa- tive-indeed, since severe COLD decreases cigarette use, they might even yield the inverse of the truth. A possible compromise might be to study the relationship between cigarette composition and the one-second Forced Expira- tory Volume (FEV) in early middle age (e.g. 35-44), before the FEV had got low enough to have much effect on smoking habits, but no recent such study has attempted this. Thus, for COLD, there is as yet no direct assessment, on an individual basis, of whether there are any important differences in the disease onset rates (i.e. in FEV loss) that are produced by different types of cigarette. In any case, even if there were, one would still not know which components of smoke were chiefly responsible. So, the above recommendations for seeking ways of producing decreases in compensation (with respect to all but a few smoke components) may be the most immediately promising means of cigarette modification to explore, for COLD as for heart disease. Changes in the amount of tobacco used Although the type of change in cigarette composition that has been introduced over the past few decades may reduce the risk of death per cigarette by a small but worthwhile amount, and although experimental investigation of the physi- cal and pharmacological determinants of compensation may lead to the design of cigarettes with still lower risks, cigarettes are still likely to kill about 20 or 30% of those who use them regularly. Effective discouragement of their use is difficult (and, in different countries, these difficulties may be quite different), but it is so uniquely worthwhile that it deserves even more attention than it gets. A variety of voluntary orgar and these deser There are, hob deficient. The fi many governm~ that may be con tax revenues, e principle goven in practice the3 government is b the tax on toba when sales are d especially becaL evidence of effe A second de! stress on the Ion ial across about : developed count admittedly diffic remembered aF framework of ot! possible, so long sages that qualfl QUARTER Ol~ KILLED BEFC less complicated groceries, car p consumer societ matter for exper Britain) be helpf SMOKING IS Among 1000 3' habout 1 will --about 6 will --about 250 wi Even in the Uni double those in E some such comp~ start with only 10 25 of them with d Whatever forn o o~ o%

PETO .ral history of COLD ends for lung cancer, aere are other impor- ,t a time when there i fferent social classes cause of death in the women. Since then, :ddle-aged men have all. It is very difficult cannot confidently ution decreases pro- e correct one. directly informative nposition on COLD . even less informa- ~e, they might even ight be to study the :ond Forced Expira- ~re the FEV had got ecent such study has • an individual basis, ;e~et rates (i.e. in • ~I~Y case, even if smoke were chiefly ways of producing ,moke components) ~tte modification to aas been introduced cigarette by a small gation of the physi- y lead to the design • to kill about 20 or .ment of their use is be quite different), re attention than it TOBACCO-RELATED DISEASE 135 A variety of WHO and UICC expert reports have been prepared on how voluntary organizations and governments can decrease cigarette consumption, and these deserve careful scrutiny, for they contain much well-judged advice. There are, however, two important respects in which they may be somewhat deficient. The first is that tax increases are not sufficiently emphasized. Because many governments derive large tax yields from tobacco sales, all the strategies that may be considered for reducing cigarette sales will also, if effective, reduce tax revenues, except for an increase in the taxation of tobacco. Although in principle governments may believe they act only for the good of their citizens, in practice they may tend to decide that what is economically easiest for the government is best for the citizens. Consequently, the one strategy--increasing the tax on tobacco---that increases rather than decreases tax revenues, even when sales are decreased, perhaps deserves more emphasis than it usually gets, especially because it is one of the few strategies for which there is clear, direct evidence of effect. A second deficiency of emphasis is that there may have been insufficient stress on the long-term advantages of getting quantitatively informative mater- ial across about the total risks from tobacco, and the extent to which, at least in developed countries, these exceed all other reliably known causes of death. It is admittedly difficult to communicate risks in a way that will be understood and remembered approximately correctly, especially by people who have no framework of other risks with which to compare them. However, this should be possible, so long as the main message is set clearly apart from the lesser mes- sages that qualify it. After all, the chief message is merely that 'ABOUT A QUARTER OF ALL REGULAR CIGARETTE SMOKERS WILL BE KILLED BEFORE THEIR TIME BY THE HABIT', which is considerably less complicated than the mass of quantitative information about house prices, groceries, car prices, etc. that already has become part of the folklore of consumer societies. How exactly this main message should be put over is a matter for experiment: comparisons with other conditions may (especially in Britain) be helpful, for example: SMOKING IS BRITAIN'S BIGGEST KILLER Among 1000 young adults who smoke cigarettes regularly, --about 1 will be murdered --about 6 will be killed on the roads --about 250 will be killed by tobacco. Even in the United States, where road accident death rates are more than double those in Britain and murder rates are about ten times those in Britain, some such comparisons may be helpful (although it may then be advisable to start with only 100 young United States adults, and to threaten about 1, 2 and 25 of them with death). Whatever format is preferred, however, the central point remains: the

I I|111111111111 II II II III IIIIIIIII III 136 PETO reason one wants to prevent smoking is not just because it is danger- ous--dozens of things are dangerous--but because it is so dangerous. This indicates getting some sort of quantitative information over, both about the effects of smoking itself on mortality and, perhaps at least as importantly, about how much smaller all reliably known other carcinogenic effects are. Such in- formation may in the short term make only a few people give up, but over a few years wide acceptance of such a perspective may have substantial effects, either on individual behaviour or on making other actions political!y acceptable. CONCLUSION Large changes in cigarette usage can be produced by socially acceptable means (Table 2). In Britain, for example, cigarette usage per adult has decreased by about 30% over the past decade alone, Likewise, tar level reductions can fairly easily be implemented, especially in countries such as Russia and China where cigarettes are manufactured and distributed by the State with little advertising and where typical tar deliveries exceed the upper limit of what is currently sold in Britain. TABLE 2 Information for governments on simple measures for the control of lung cancera Price increases will produce fewer deaths and more revenue (as long as the)' do not feed back into wage demands) Tar reductions should be encouraged (especially in countries such as Russia and China where typical tar levels are still of the order of 20-30 mg, which is extremely high) Advertising could be taxed, restricted, prohibited, or limited to cigarettes delivering under I0 mg tar Simple, clear, quantitative information could be communicated effectively ~o the general population: ABOUT A QUARTER OF ALL REGULAR SMOKERS ARE KILLED BEFORE THEIR TIME BY TOBACCO I General note: recommendation of these few simple measures (which might have a substantial effect in just a few years on exposure) does not, of course, detract from the need for a wide range of other measures, including many of those suggested by WHO (1979), UICC (Gray 1977), and others. Modificztions of the consumption and of the composition of cigarettes are complementary, not competing, strategies that together could lead to avoidance of most of the 100 000 or more tobacco-induced deaths each year in Britain, most of the 300000 or more in the US, and some hundreds of thousands in the rest of the world. Such changes will, perhaps surprisingly, TOBACCO-REL not greatly cha developed cou~ in middle age. REFERENCE~ Cederl6f R. Doll assessment meth Doll R. Peto R It. doctors. Br Med Doll R. Peto R 19~ United States to,_ Gray N (ed) 1977 1. Contre le Cancer Independent Scien Stationery Offic~ Kaufman DW. He monoxide conter 308:409-413 Lee PN 1976 Stat~ Research Counc~ Lee PN, Garfinket 35:16-22 Peto R. Speizer FE Richards SM. Gi but not of mucus 128:491-500 Wald N, Idle M, B cigarette. Thorax Winn DM, Blot cancer among w~ World Health Org~ committee on sm DISCUSSION Doll: I don't ' disease in Brita diagnostic habit 20 or 30 years a~ term. The diff, differences, bec Koplan: You obstructive lung

PETO ~cause it is danger- ~ so dangerous. This ~ver, both about the as importantly, about effects are. Such in- ve up, but over a few tantial effects, either =ally acceptable. tlly acceptable means lult has decreased by reductions can fairly sin and China where ,'ith little advertising .'hat is currently sold I of lung cancer~ ~ot feed back into ~ussia and China where s delivering under 10 mg ectively to the general LED BEFORE THEIR night have a substantial need for a wide range of .~ICC (Gray 1977), and 9n of cigarettes are ~er could lead to deaths each year in some hundreds of rhaps surprisingly, TOBACCO-RELATED DISEASE 137 not greatly change the life expectancy of old people, but they will, at least in developed countries, appreciably decrease the proportion of people who die in middle age. REFERENCES Cederlrf R. Doll R. Fowler B, Friberg L, Nelson N, Vouk V 1978 Air pollution and cancer: risk assessment methodology and epidemiological evidence. Environ Health Perspect 22:1-12 Doll R. Peto R 1976 Mortality in relation to smoking: 20 years' observations on male British doctors. Br Med J 2:1525-1536 Doll R, Peto R 1981 The causes of cancer: quantitative estimates of avoidable risks of cancer in the United States today. J Natl Cancer Inst 66:1191-1309 Gray N (ed) 1977 Lung cancer prevention: guidelines for smoking control. Union Internationale Contre le Cancer, Geneva Independent Scientific Committee on Smoking & Health 1983 Third Report. Her Majesty's Stationer3' Office. London Kaufman DW. Helmrich SP, Rosenberg L, Miettinen OS, Shapiro S 1983 Nicotine and carbon monoxide content of smoke and the risk of myocardial infarction in young men. N Engl J Meal 308:409-413 Lee PN 1976 Statistics of smoking in the United Kingdom. Research Paper No. 1. Tobacco Research Council, London Lee PN, Garfinkel L 1981 Mortality and type of cigarette smoked. J Epidemiol Community Health 35:16-22 ~ Peto R, Speizer FE, Cochrane AL, Moore F, Fletcher CM, Tinker CM, Higgins ITr, Gray RG, Richards SM, Gilliland J, Norman-Smith B 1983 The relevance in adults of airflow obstruction. but not of mucus hypersecretion, to mortality from chronic tung disease. Am Rev Respir Dis 128:491-500 Wald N, Idle M, Boreham J, Bailey A 1980 Inhaling habits among smokers of different types of cigarette. Thorax 35:925-928 Winn DM, Blot WJ. Shy CM. Pickle LM, Toledo A, Fraumeni J 1981 Snuff-dipping and oral cancer among women in the Southern United States. N Engt J Med 304:745-749 World Health Organization 1979 Controlling the smoking epidemic: report of the WHO expert committee on smoking control. Technical Report Series 636. WHO, Geneva DISCUSSION Doll: I don't think one can really compare trends in chronic obstructive lung disease in Britain and the US. There were formerly enormous differences in diagnostic habits between the two countries, which have diminished in the last 20 or 30 years as a result of physicians getting together and deciding to use this term. The differences in trends can be largely attributable to nosological differences, because the national statistics have started from a different base. Koplan: You say that there are other aetiological factors involved in chronic obstructive lung disease (COLD), in addition to smoking. In lung cancer itself,

138 DISCUSSION you suggest that one means of comparing low tar cigarette smokers to high tar sm6kers is to compare data over time. Tar content clearly is the major compo- nent in lung cancer, but are there likely to be other aetiological agents for lung cancer as well, in parallel to COLD? Peto: There certainly are other host factors in lung cancer. There are obvious ones, like asbestos and ionizing radiation. It is also possible that there are nutritional components in lung cancer, or that infective processes could be relevant. I don't think changes in air pollution can be plausibly invoked as causes of the changes over time. partly because air pollution was never an important cause of lung cancer anyway, and partly because one sees the same trends in Finland, a country that has never been seriously polluted (outside the sauna baths!). Blanpain: The pine woods in Finland produce substantial air pollution: substances such as turpentine are turned into carcinogenic smog under the influence of sunlight. Peto: Changes in air pollution from pine forests are not responsible for the 40% decrease seen in Finnish lung cancer rates between the early 1960s and the late 1970s. The number of cigarettes per man smoked in Finland, like that in Britain, stayed roughly constant until the mid 1970s and only then began to fall, with large price increases. The tar deliveries fell enormously, because there was a switch from Russian-type cigarettes to modern Western low tar cigarettes. In fact, the fall in tar delivery in Finland was greater than anywhere else in the world, and we saw a 40% decrease in lung cancer incidence in early middle age by the late 1970s. By now, the decrease may be beyond 50%, as it is in Britain. Miteva-Toncheva: We are confronted here with a very complex problem, namely the interaction between environmental factors and the individual's genotype (for example, the effects of tobacco on potentially susceptible indi- viduals). Have you investigated al-antitrypsin variants by isoelectrofocusing? Homozygotes for deficiency of this protease inhibitor, and heterozygous car- riers, are highly susceptible to chronic obstructive lung disease (COLD) under the influence of tobacco. Such individuals form a high risk group, and it is very important to give more care to them. Peto: For homozygotes, that is true. People deficient in this inhibitor finish up with proteases coming out of their alveoli, so wrecking their lungs complete- ly, and may die of COLD at the age of 40. Homozygotes with al-antitrypsin deficiency are rare and are at enormous risk of this lung disease, if they smoke. Heterozygotes are an appreciable proportion of the population and some studies have suggested that they are at increased risk but, overall, there seems to be no substantial difference in risk between heterozygotes and people of normal genotype. But there must be genetic deter)ninants of the response to cigarette smoke of the heart, lungs, and other systems. However, we can change the environment, but we can't change the genotype. TOBA Met smoki peopl~ ischae decre: Pet, could variot, On smoki and ht switch cigare moutl- don't becau their i~ that ci manta in the chang~ Hje., (CHD becau, the 0, sclero, reasor Pet~ CHD so extz from .~ fivefol high a Sec~ extren diseas, causal Dan smokil only c, (Fribe" young, 0

DISCUSSION okers to high tar ~e major compo- agents for lung here are obvious e that there are ,cesses could be ~ibly invoked as n was never an ~e sees the same ~ted (outside the air pollution: ~mog under the responsible for early 1960s and and, like that in _~n began to fall, cause there was ir~rettes. In ~glse in the arI~middle age it is in Britain. ~plex problem, he individual's ~sceptible indi- ectrofocusing? erozygous car- COLD) under ~, and it is very nhibitor finish tngs complete- al-antitrypsin if they smoke. on and some 1, there seems md people of e response to ever, we can TOBACCO-RELATED DISEASE 139 Meade: What is the current status of the evidence on cigars in relation to smoking-related disease, cardiovascular or malignant, now that so many more people are smoking them? You also mentioned the high relative risk of ischaemic heart disease in younger cigarette smokers, and the fact that it decreases with increasing age. Have you any ideas why that is? Peto: I don't know why the relative risk changeswith age like that. You could produce models in which you eliminate susceptible sub-groups in various ways, but that is a mathematical exercise, not a biological one. On the question of cigars, one has to distinguish between two types of cigar smoking. The first is in people who smoked cigars when they began smoking and have smoked them throughout their lives. There are also people who have switched from cigarettes to cigars because they want to avoid the health risks of cigarettes. Long-term cigar smoking confers a certain risk of cancer of the mouth and throat, but nothing like the lung cancer risk of cigarettes. But we don't know what will happen when people switch from cigarettes to cigars, because this could produce quite different effects. These people may carry over their inhaling patterns to cigars. There is not much evidence, in animals at least, that cigar smoke is less hazardous than cigarette smoke. It is perhaps chiefly the manner in which the cigar is smoked that is protective, rather than differences in the make-up of the smoke, except insofar as these differences produce changes in the manner of smoking. Hjermann: I believe that smoking is causally linked to coronary heart disease (CHD), but I am not quite sure how firmly I should believe this! It is partly because of the lack of controlled trials in this field. In a prospective study like the Oslo trial, we could not find any significant correlation between athero- sclerosis and smoking. It seems to be hard to show. What are your main reasons for believing in this causal relationship? Peto: The fundamental reason for believing that the relationship between CHD and smoking is a causal one is that the relative risk in early middle age is so extreme. Among people in their thirties, CHD is rare but the relative risk from smoking is as much as 10-fold. In the 40-44 age group there is still a fivefold relative risk. So the relative risk between the ages of 30 and 50 is so high as to make non-causal explanations rather implausible. Second, peripheral vascular disease and aortic aneurysm have even more extreme relative risks among smol~ers than CHD does. For peripheral vascular disease the association is so extreme as v~,rtually to preclude anything except a causal explanation. Danielsson: There is still some question about the causal correlation between smoking and ischaemic heart disease, in my view. A twin study showed that the only certain correlation between smoking and disease was for lung cancer (Friberg et al 1973). The problem with evidence for causality dra~vn from the younger smokers is that you don't have a good control group. We should always

-- - IlII lilllllll ..... I III , ............. ,, ,,,, , ........... 140 DISCUSSION TOBACCO-F bear in mind that perhaps a smoker is different from a non-smoker; that is, smokers may be more prone to this disease, whether or not they actually smoke. This is why it is so difficult to do a conclusive study. Peto: This is not a plausible hypothesis where the relative risks are so extreme. The twin studies are much overrated, Like randomized trials, they would be fine if done on a scale one or two orders of magnitude bigger than they • .are done. When the evidence from homozygous and heterozygous twins is studied, and one looks at the actual numbers of events and how much differ- ence there was in smoking, there is not enough evidence to add usefully to that from much larger studies. Hiatt: Is any effect on cancers other than those in the respiratory tract visible yet, with the changes in cigarette smoking? And what are your present thoughts about interactions between cigarettes and other carcinogens? Peto: We can't really assess the effects of changes in cigarette composition on say, cancer of the bladder, although Paolo Vineis has suggested, on the basis of his large but unpublished case-control study, that black tobacco (like that in Gauloises, for example) is peculiarly hazardous. One would probably learn more by measuring the extent to which the urine of smokers is mutagenic when they are smoking different types of cigarette, taking the mutagenicity of the urine as a surrogate for epidemiology. Some types of cigarettes deposit sub- stantial amounts of benzidine into the bladder. The chief source of benzidine that people are now exposed to in Britain is from cigarettes, rather than the industrial use of dyes and other chemicals. Doll: Mortality from cancer of the bladder and of the pancreas---diseases that are not very strongly related to tobacco use--is either falling or static in the UK. However, the incidence rates of some cancers that are closely related to tobacco--cancer of the mouth, oesophagus and larynx--are going up. These last changes presumably result from the increase in alcohol consumption, which is even more strongly related to these cancers. These changes illustrate the complexity of drawing conclusions about the effect of specific aetiological agents from mass statistics. If I wanted to argue that cigarette smoking was not proved to be the cause of any type of cancer, I would point to the trends in cancer of the larynx in various countries. Although this cancer has been closely related to cigarette smoking in all case-control studies, the trends in incidence and mortality have sometimes been in the opposite direction to the trends for cancer of the lung. To my mind, the only reasonable explanation is that other factors of which we are still ignorant account for these exceptions. But if you want to allow an exception to disprove a massive set of evidence, you point to the trends in cancer of the larynx. These cases are important to investigate because they are exceptions, but we ought to beware of allowing such exceptions to undermine a mass of contrary evidence, when we ar and the pro. Shephara naive to thi into sugges other risks will make i With reg obtained bx In fact, wit cigarettes, Peto: Th airways. I a the alveo/i reductions, young peoI continue t~ tendency tt levels--the levels were when Japm took up th~ that high t~ Koplan: cigarette s~ non-tobacc Peto: I d, between cc The comm public pres really seri~ European efforts; it c health, or problem s~ where the serious, an reduction, years, whi, Eddy: V~ of tobacco

DISCUSSION • moker; that is, ~t they actually ve risks are so zed trials, they qgger than they zygous twins is ,w much differ- usefully to that ory tract visible ,resent thoughts composition on • on the basis of co (like that in probably learn ~utagenic when genicity of the es deposit sub- ze of benzidine ,, or static in the ~sely related to ~ing up. These consumption, ions about the anted to argue pe of cancer, I ries. Although 11 case-control .~s been in the mind, the only • still ignorant on to disprove • larynx. These ,ut we ought to rary evidence, TOBACCO-RELATED DISEASE 141 when we are trying to interpret the relationship between aetiological agents and the production of disease. Shephard: You discussed the tactic of emphasizing tar reduction. It may be naive to think that cigarette manufacturers are not going to pervert this view into suggestions that we believe low tar cigarettes to be safe. There may be other risks with this tactic, one being that the lower levels of tar and nicotine will make it easier for young people to become addicted to cigarettes. With regard to the figures for the tar content of cigarettes, these are all obtained by smoking machines which presuppose a certain pattern of smoking. In fact, with the change in pattern of smoking that one gets with low tar cigarettes, there may be very little reduction in the tar delivered by them. Peto: There may be a substantial difference in tar delivery to the main airways. I agree that there may be no great change in the amount that reaches the alveoli. However, in countries where there have been good tar level reductions, there have also been considerable reductions in the extent to which young people are taking up smoking and also in the extent to which adults continue to smoke. That is not always true. but at least there is no systematic tendency the other way. Conversely, consider countries with very high tar levels--the USSR at present, or Japan in the 1950s. At that time in Japan, tar levels were high. Smoking by minors was forbidden during those years, but when Japanese males reached adulthood, about 80% of the male population took up the smoking of high tar cigarettes. So there is no suggestion, either, that high tar levels discourage people from starting to smoke. Koplan: Are there any differences in the level or type of effort to curtail cigarette smoking among countries that are tobacco producers, as against non-tobacco producers? Peto: I don't really know. There does, however, appear to be a big difference between countries where tobacco manufacture is in private and in public hands. The communist countries have been terrible; I suppose there is no effective public pressure on those governments to act, and so they have done nothing really serious to discourage smoking in the USSR, China or any Eastern European country. In at least some Western countries there have been serious efforts; it depends very much on concerned individuals within departments of health, or within the population generally. One or two people who take the problem seriously can make a vast difference to a whole country. In France, where the State manufactures tobacco, the government has done nothing serious, and in Britain it has. Perhaps partly because of this, Britain has seen a reduction of one-third in the number of cigarettes smoked over the last i0 years, which is a considerable success that should be emulated elsewhere. Eddy: What role do you think nicotine chewing gum might play in the control of tobacco-related diseases?

142 DISCUSSION Peto: A relatively minor one. Various randomized trials have suggested some effect, though not a large one. But, because the problem is so important, a minor effect on a major disease represents a useful public health advance, so it is a worthwhile aid that should be introdu_ced, encouraged and used. Any- thing that helps to get people off cigarettes is a good idea. We have to find out what works for different people, and nicotine chewing gum helps a few people. REFERENCE Friberg L, Cederl6f R, Lodch U et al 1973 Mortality in twins in relation to smoking habits and alcohol problems. Arch Environ Health 27:294-304 Screenir related Malm6 " BO PETERSSOt' Section of Preve: S-214 O1 MalmO, Abstract. related di: in a large underlyin! In a pr~ death in n most pred cant corre tively, ant In the r admissior group, af" prcventir related di as the ret high/non 1985 The p 143-16 The preventi broad progrz as individual complete eli~ Three mai phasized, na subject, and The purp~ individual st.