Tobacco Documents Online

155 LUNG CANCER, CORONARY HEART DISEASE AND SMOKING By H. J. Eysenck, Ph.D., D.Sc. Professor of Psychology Institute of Psychiatry University of London March 10, 1982 0 ~.? ~ 0 ~ ~ ~ 9i-n77 O-k1--11 ~

156 Statement of Professor Hans J. Eysenck I am Hans J. Eysenck, professor of psychology at the Institute of Psychiatry, University of London and psychologist to the Maudsley and Bethlem Royal hospitals in London. I received my Ph.D. in 1940 and my D.Sc. in 1964, both from the University of London. I was Senior Research Psycho- logist at Mill Hill Emergency Hospital from 1942 through 1946. In 1949 and 1950 I was a visiting professor at the University of Pennsylvania in Philadelphia. Between 1950 and 1954, I was a Reader in Psychology at the University of London's Institute of Psychiatry. In 1954 I was a visiting professor at the University of California at Berkeley. I am a Fellow of both the British Psychological Society and of the American Psychological Association. I have founded and edited three psychological journals, and I am on the editorial boards of some 15 other inter- national psychological journals. I have written or edited for publication approximately 35 technical books and over 600 articles dealing with various aspects of the psychological field, particularly with respect to personality, intelligence, 157 behaviour therapy and behavioural gene research in the area of smoking for ov authored two books, the most recent of The Causes and Effects of Smoking, as articles on this subject. A widely accepted theory asserts causes lung cancer, coronary heart dis diseases with which it is statisticall always realized that (a) such a theory and is beset by many anomalies and dou there is an alternative theory which 1 facts which are not explained by the c present position seems to be that eith the tragic incidence of lung cancer an disease (to which this brief account w that both may be needed to complement There is agreement that smoking nor a sufficient cause of lung cancer. less than 10 will develop lung cancer; a sufficient cause. And of 100 peoplE cancer, approximately 10 will be non-: is not a necessary cause. This simple numbers differ of course from country -2-

157 behaviour therapy and behavioural genetics. I have conducted research in the area of smoking for over 20 years and have authored two books, the most recent of which is entitled The Causes and Effects of Smoking, as well as numerous articles on this subject. A widely accepted theory asserts that cigarette smoking causes lung cancer, coronary heart disease, and many other diseases with which it is statistically linked. It is not always realized that (a) such a theory is far from proven, and is beset by many anomalies and doubts, and that (b) there is an alternative theory which is based on undeniable facts which are not explained by the causal theory. The present position seems to be-that either theory may explain the tragic incidence of lung cancer and coronary heart disease (to which this brief account will be restricted), or that both may be needed to complement each other. There is agreement that smoking is neither a necessary nor a sufficient cause of lung cancer. Of 100 heavy smokers, less than 10 will develop lung cancer; hence smoking i8 not a sufficient cause. And of 100 people who develop lung cancer, approximately 10 will be non-smokers: hence smoking is not a necessary cause. This simple fact (the precise numbers differ of course from country to country, but indicate -2-

158 the correct order of magnitude) suggests that the scientific proof for any particular theory will be difficult to arrive at, and that any such theory will almost certainly be complex and multi-faceted. Much of the evidence cited in favour of the causal theory is statistical, but many statisticians have severely criticized the evidence on statistical grounds. Such suggested proofs as the correlation between smoking and lung cancer within a given country, or between lung cancer and number of cigarettes smoked between countries, are evidence of correlation, not of causation; one of the first lessons the budding statistician learns is that correlation does not imply causation. (There is a very high correlation between countries linking meat eating and cancer of the large intestine, yet we do not conclude that eating meat causes cancer of the large intestine!). Hence this method of demonstration, while suggestive, is far from compelling. This would be so even if the figures usually quoted could be taken seriously; however, there are good reasons for doubting their accuracy. The figures quoted are based on clinical diagnosis of lung cancer, but these are very unreliable and imprecise. If we take as our criterion autopsy data, and compare these 159 159 with routine diagnosis, we find that pr= out of 100 people found on autopsy to h: cancer, only 3 were so diagnosed. This very obvious under-diagnosis of lung car In recent years, exactly the opposite ha an over-diagnosis of lung cancer of up t Whether these changes in diagnostic pref responsible for the alleged tremendous i over the years or not, and whether it ma the observed correlation between lung ca is impossible to say; all we can say is data so completely unreliable, the stati. are suspect. Another important point concerns ti from other, correlated habits, such as d: up, staying out late, wenching, etc., i.c life the totality of which may increase t so that smokers are biologically older tt given age, for reasons only partly invol~ Non-smokers are different types of persor generally more self-protective, and the F and habits thus linked with non-smoking ¢ to the longevity of non-smokers than thei smoke. , -4- -3-

y 159 with routine diagnosis, we find that prior to World War 1, out of 100 people found on autopsy to have died of lung cancer, only 3 were so diagnosed. This is typical of the very obvious under-diagnosis of lung cancer then prevalent. In recent years, exactly the opposite has been found, namely an over-diagnosis of lung cancer of up to 200X and more! Whether these changes in diagnostic preference are completely responsible for the alleged tremendous increase in lung cancer over the years or not, and whether it may in part account for the observed correlation between lung cancer and smoking, it is impossible to say; all we can say is that arith the basic data so completely unreliable, the statistics based on them are suspect. Another important point concerns the isolation of smoking from other, correlated habits, such as drinking, living it up, staying out late, wenching, etc., i.e. a certain style of life the totality of which may increase the "rate of living", , so that smokers are biologically older than non-smokers at a given age, for reasons only partly involved with smoking. Non-smokers are different types of persons from smokers, are generally more self-protective, and the personality traits and habits thus linked with non-smoking may be more relevant to the longevity of non-smokers than their refusal to smoke. -4-

160 It is often suggested that sex differences, with males showing more lung cancer, are the product of the tendency of males in the past 50 years or so to smoke more. However, as several authorities whom I quote in my book have pointed out, similar sex ratios to those observed now were found before cigarette smoking became popular. Again, it is found that changes in the rate of increase of lung cancer diagnosis occurred simultaneously for men and women, although the women, who took up smoking much later than men, should have shown these changes at a much later date than men. If the causal theory is true, then we would expect a definite dose-response relationship; in other words, the heavy smoker should be light smoker. Yet the stricken with cancer earlier than the amount smoked makes no appreciable difference to the mean age at which the person is reported first to the clinic. Again, inhalation should make lung cancer much more likely than smoking without inhaling, yet the figures show if anything an opposite trend. These two observations are difficult to reconcile with the theory of smoking. causal The most impressive evidence for the causal theory has been the report that physicians who gave up smoking showed less lung cancer than members of the general public who 161 continued to smoke. Thus, it might ap; smoking has saved the lives of those w: proof is only acceptable if those who and those who later on give up smoking identical with respect to their health gave up smoking. Clearly, if those wb smoking are already much healthier tha continue to smoke, then the final diff be due to the already existing differe up smoking, rather than to the cessati there is good evidence to show that s¢ already differed with respect to theii the ex-smokers gave up smoking. Simi~ that from the point of view of person; different from continuing smokers. T' is based on an erroneous assumption. These objections tQ the causal in my book, do not prove the theory t argue that it is still only a theory, More convincing proof is required bef accorded a more advanced status. But there are numerous facts suggesting : and these facts cannot easily be int- theory. Yet a proper theory demands to all relevant facts, and thus agai found wanting. -5- -6-

161 continued to smoke. Thus, it might appear that giving up smoking has saved the lives of those who did so. But this proof is only acceptable if those who continue to smoke, and those who later on give up smoking, are essentially identical with respect to their health before some of them gave up smoking. Clearly, if those who later on give up smoking are already much healthier than those who later on continue to smoke, then the final differences in health may be due to the already existing differences before anyone gave up smoking, rather than to the cessation of this habit! But there is good evidence to show that smokers and ex-smokers already differed with respect to their health record before the ex-smokers gave up smoking. Similarly, there is evidence that from the point of view of personality ex-smokers are different from continuing smokers. Thus this alleged proof is based on an erroneous assumption. These objections tQ the causal theory, and others made in my book, do not prove the theory to be wrong; they simply argue that it is still only a theory, not a scientific law. More convincing proof is required before the theory can be accorded a more advanced status. But further than that, there are numerous facts suggesting an alternative theory, and these facts cannot easily be integrated with the causal theory. Yet a proper theory demands that attention be paid to all relevant facts, and thus again the causal theory is found wanting. -6- ®

162 The alternative theory, first suggested by the eminent geneticist and statistician Sir Ronald Fisher, suggests that genetic factors are important in causing lung cancer; that genetic factors are active in causing people to maintain the smoking habit; and that possibly the same genetic factors may be involved in both these trends, thus producing the observed correlation between smoking and cancer (insofar as'such a correlation is real). There is evidence that genetic factors do play a part in the causation of lung cancer; this is not in doubt. I have brought forward evidence (in addition to already very convincing evidence produced by many other people) to show that genetic factors are relevant to the maintenance of the smoking habit. Thus there is evidence for both the assumptions on which Fisher's argument was based. The origin of the smoking habit, on the other hand, is hardly at all influenced by genetic factors. It appears from our genetic analysis and from the direct study of the problem by Professor Spielberger that the origin of the smoking habit is due to peer pressure; parental influences play a much smaller part, and advertising almost none. My own contribution has been to suggest that the mediat- ing factor between cancer and smoking may be the personality of the people involved. Thus it is assumed that people of 163 a certain personality are more likely lung cancer irrespective of smoking. people of a certain personality are mc others. There is evidence for both t: original work with Dr. Kissen, an emir showed very marked personality differF patients and patients suffering from r with the personality assessment made t then, a large-scale study in East Gerrr findings (themselves replicated in anc and has found similar personality tra' of lung cancer patients in women with Other studies, also indicating a rela* and personality, are cited in my book. In a similar way, my early work established a correlation between per~ and many studies in different countri~ our findings, and added new ones. We fundamental assumption of Fisher's ge empirical support, and we may add tha modest support for my own attempt to major fields. Unfortunately there ha along these unusual and somewhat unor -8- -7-

163 a certain personality are more likely than others to die of lung cancer irrespective of smoking. It is also assumed that people of a certain personality are more likely to smoke than others. There is evidence for both these propositions. My original work with Dr. Kissen, an eminent British oncologist, showed very marked personality differences between lung cancer patients and patients suffering from non-malignant tumours, with the personality assessment made before diagnosis. Since then, a large-scale study in East Germany has replicated our findings (themselves replicated in another study by Kissen), and has found similar personality traits to those characteristic of lung cancer patients in women with cancer of the breast. Other studies, also indicating a relation between lung cancer and personality, are cited in my book. In a similar way, my early work with Tarrant and Woolf established a correlation between personality and smoking, and many studies in different countries have since confirmed our findings, and added new ones. We may thus say that the fundamental assumption of Fisher's genetic theory have found empirical support, and we may add that there is also some modest support for my own attempt to integrate these two major fields. Unfortunately there has been too little work along these unusual and somewhat unorthodox lines to say that -8-

164 the results are anything more than suggestive, and the theory linking them is still in a very elementary stage; nevertheless, as far as the findings go they support the genetic rather than the causal theory, although they do not necessarily contradict the latter. It seems unfortunate that the premature crystal- lization of spurious orthodoxies has prevented the genetic theory from attracting sufficient research grants to work it out in sufficient detail, and to carry out the research necessary to put it on a more acceptable footing. Recently some progress has been made on the theoretical development of the genetic hypothesis by linking it with research on stress, in particular the differential effects of chronic and acute stress, and the "inoculation" theory of stress. However, in the absence of large-scale research into the refinements of this theory, and more widespread familiarity with and criticisms of its details, not too much should be claimed for it other than it presents a viable alternative to the causal theory. In relation to the causal theories of coronary heart disease (CHD), similar criticisms apply as do in the case of lung cancer. There are considerable unreliabilities in diagnosis; there are large numbers of factors other than smoking which have been associated and which are not usually -9- 165 controlled for in studies of the effec- do not on the whole differ from non-in. proneness; the statistical relation be and CFD disappears in many countries, Yugoslavia, Italy, Greece and Japan; t dose-response relationship, i.e. there between duration of heavy cigarette sm myocardial infarction; and the correla cigarettes smoked and CHD is not linea studies appear to be safer than non-sm CHD, such as angina pectoris (which co CfID in men) fail to show even a statis cigarette smoking; some types of smoki to show even a statistical correlatior are anomalies or failures of the causz an explanation before the causal theoi Some of these facts are much more reac of a genetic-personality theory; thus of cigarette vs. pipe/cigar smoking m: in terms of the known differences in I associated with these different smoki: The general conclusion would se case of CHD, as in the case of lung c causal influence of smoking is still -10-