Tobacco Documents Online

7.2 September 1983 Risk Factors for Coronary Heart Disease Presented by Dr. George Schafer A gambler knows that his chance (risk) of rolling a 7 with a pair of dice is one in six. On the first roll he will win if a 7 or an 11 appears, and lose if a 2, 3, or 12 appear. His chance of winning is eight in thirty-six, and his chance of losing is four in thirty-six. The odds (relative risk) that he will win rather than lose on the first roll are eight-to-four or two-to-one. This concept of odds was related to disease entities by Woolf (1955) and Cornfield (1956). Woolf demonstrated a numerical technique for estimating "incidence ratios" of some diseases in relation to blood groups. He calculated "odds" that a disease (peptic ulcer) would occur among persons in blood group 0 as compared with blood group A. Cornfield, adopting the term "relative risk", presented a similar numerical technique for calculating the odds that death from lung cancer would occur in a cigarette smoker as compared with a nonsmoker.

At the time of the U.S. Terry Report (1964) on Smoking and Health, the term "relative risk" did not have wide usage. Instead, an analogous term, "mortality ratio", was used to compare the risks of death from a specific disease in various groups. In this way, the risk of death, or mortality rate for smokers was related to an estimated mortality rate for nonsmokers. Where the incidence or the prevalence of a disease (rather than mortality) was considered, the term "risk ratio" was applied. The resulting statistics, namely "mortality ratio" and "risk ratio", were, for all intents and purposes, "relative risks". In 1964, the Terry Report relied principally on the results of seven prospective studies to report its conclusions concerning mortality ratios for smokers dying from various diseases. Reports of three of these seven studies have appeared repeatedly in the medical and scientific literature and in subsequent Surgeon General Reports during the past nineteen years. These are (1) The Doll-Hill study of British doctors (1964), (2) The Dorn-Kahn study of U.S. Veterans (1966), and (3) The American Cancer Society studies by Hammond of U.S. adults (1958, 1966). On the subject of coronary heart disease(CHD), the Surgeon General's Committee reported in 1964 a mortality ratio of 1.7 for cigarette smoking males. It concluded that, "Male cigarette -2-

smokers have a higher death rate from coronary artery disease than non-smoking males, but it is not clear that the association has causal significance." Further results on CHD based on the American Cancer Society "one-million person" study were reported by Hammond (1966). A Public Health Service Review (1967) summarized these as follows: Smokers between ages 45 and 54 have the highest mortality ratios (relative risks) as compared with non-smokers - three times as great for men, and twice as great for women who smoke ten or more cigarettes per day. The same PHS Review discussed the concept of "multiple risk factors" that affect CHD incidence and mortality. It reported the seminal treatise by Truett, Cornfield, and Kannel (1967) which had presented computations based on data compiled in the long-term study of heart disease in Framingham, Massachusetts. This research led the authors to conclude that "the most on important risk factors, aside from age itself, are cholesterol, cigarette smoking, ECG abnormality, and blood pressure." The Framingham study thus became the frame of reference from which the concept of "risk factors" was established as dogma in medical and epidemiological considerations of CHD mortality and morbidity in the 1970's. Results of prospective studies dealing with smoking and CHD mortality ratios were summarized in the 1971 report to the -3-

Surgeon General. In general, these mortality ratios did not deviate from the median of 1.7 and range of 1.5-2.0 cited in the 1964 Report. The identical summary tables were cited in the 1976 Report and again, without embellishment, in the 1979 Report to the Surgeon General. Thus, the CHD mortality ratio (relative risk) for smokers that was estimated from the earliest epidemiological studies has not been altered in the two decades since these results were first reported. From the outset, questions were raised in the scientific literature concerning the adequacy of these epidemiological studies, and their suitability as evidence upon which sweeping conclusions could be drawn. In the forefront of the critical attack was a small, but eminent group of statisticians that included Fisher (1957,1959), Berkson (1955,1958,1959), Brownlee (1965) and Katz (1967). All agreed with the assertion of the Terry Report that "statistical methods cannot establish proof of a causal relationship in an association". At the same time each recognized the underlying failure of the Report to pursue all the logical possibilities inherent in an association between smoking and disease. Among their more poignant criticisms was the failure of the sampled individuals to adequately represent the population to which the conclusions were being projected. ~ OS -4- W

In a review of evidence through 1967 on CHD, Seltzer (1968) concluded that it is "difficult to see from the new epidemiological data how valid causal inferences can be drawn that cigarette smoking is linked to excess CHD deaths or that the excess CHD deaths are 'caused by' cigarette smoking." Two years later, Seltzer (1970) wrote, "The statistical association of higher mortality from CHD in cigarette smokers still remains to be explained." Results of the British doctor study, one linchpin of the 1964 Report, were criticized by Seltzer (1972) for their inconsistencies in the patterns of coronary heart disease. In this appraisal of a Royal College of Physicians' Report (1971), Seltzer showed that death rates from CHD declined in the early years of the study and rose later, whereas death rates from all cardiovascular diseases rose "substantially" at first and declined "substantially" in later years. A lengthy comment to the editor by Seltzer (1975) pointed to the uncertainty of smoking-cessation intervention on "cardiac efficacy". Seltzer took this opportunity to criticize the Multiple Risk Factor Intervention Trial (MRFIT) which was then underway and correctly anticipated the difficulty this study would have in disentangling the "effects due to cessation of smoking" f rom "the effects of the confounding variables." With this in mind he observed that, "If smoking is related to CHD in only a limited segment of the -5-

population, the people who are not at risk will hardly be benefited by blunderbuss interventions aimed at everyone." A recent paper by Seltzer (1980) arrived at the conclusion "it is reasonable to believe that stopping smoking does not reduce the risk of CHD, and that there is no established proof that cigarette smoking is causally related to coronary heart disease." Support for Seltzer's conclusions appears in contem- poraneous reports from England. Burch (1979,1980) examined the secular increases and age-dependence of recorded death rates from CHD and concluded that these could not be reconciled with the view that cigarette smoking actually causes the disease. The evidence, in Burch's opinion, indicates that for cigarette smoking and other "risk factors", the observed associations with CHD have a genetic basis. His assessment of the existing data led him to the conclude that intervention programs "are not only doomed to disappointment, they are likely to create unpecessary misery and anxiety". These views have been supported by the results of the seven-year Multiple Risk Factor Intervention Trial in the United States (1982). This massive study appears to have contradicted conventional wisdom about "risk factors" for heart disease. Moreover, the MRFIT Research Group raised the anxiety levels of members of the medical profession by suggesting that anti-hyper- tensive drugs that have been prescribed extensively for more -6-

than a decade might not be effective in reducing mortality from heart disease. MRFIT's surprise results, and the inconclusive- ness of such intervention trials in general have been the subject of editorials, letters, and articles in the Journal of the American Medical Association, The Lancet, British Medical Journal, Medical Tribune, Science, American Journal of Epidemiology and other principal medical and scientific journals around the world. Following is a sampling from some of these commentaries: "Unfortunately, the fundamental question facing the investi= gators at the beginning of the experiment remains unanswered." Lundberg (1982) "So, on the face of it, it looks like risk factor reduction may not be beneficial, contrary the to current medical dogma." Kolata (1982) "The results prove nothing, and we must turn elsewhere to answer the question, Does prevention work?" The Lancet (1982) "My conclusion is that the best explanation for the failure to detect a beneficial effect in MRFIT is that no benefits accrued." Stallones (1983) -7-

"It is to be hoped that the intense preoccupation witn con- ventional risk factors will now diminish ana that the search for the genuine precipitating agent(s) might commence." Burch (1983) "It is, therefore, high time for epidemiologists and others to reexamine the basic validity of the risk factors themselves in affecting the rate of CHD mortality." Seltzer (1983) "The inevitable question remains: Do the antihypertensive drugs used in all these trials have adverse metabolic effects that outweigh the beneficial effect of BP lowering?" Leren et al. (1983) In commenting on the epidemiological evidence in 1964 the Surgeon General's committee acknowledged its unwillingness to assert that the observed association between smoking and coronary heart disease had "causal significance". Yet by 1979, in spite of only minor changes in the nature of the epidemiological evidence, the Surgeon General was willing to conclude, "that smoking is causally related to coronary heart disease for both men and women in the United States." The failure to consider in full the evidence contrary to the majority point of view has led to statements concerning CHD are manifestly biased and subject to serious scientific challenge. that -8-

REFERENCES 1. Berkson, J. (1955), "The statistical study of association between smoking and lung cancer", Proc Mayo Clinic, Vol. 30, pp. 319-348. 2. Woolf, B. (1955), "On estimating the relation between blood group and disease", Ann Human Genetics, Vol. 19, pp. 251-253. 3. Cornfield, J. (1956), "A statistical problem arising from retrospective studies", Proceedings of the Third Berkele S m osium on Mathematical Statistics and Probability, Vol. , pp. 135-1 . 4. Fisher, R.A. (1957), "Dangers of cigarette smoking", Br Med J, Vol. 2, pp. 297-298. 5. Berkson, J. (1958), "Smoking and lung cancer: Some observations on two recent reports", J Amer Stat Assoc., Vol`. 53, pp. 28-38. 6. Hammond, E.C., Horn, D. (1958), "Smoking and death rates - report of 44 months of follow-up of 187,783 men", I. Total Mortality, II. Death rates by cause, J. Amer Med Assoc. 166(11): pp. 1159-1172 and pp. 129 -130 . 7. Berkson, J. (1959), "The statistical investigations of smoking and cancer of the lung", Proc Mayo Clinic, Vol. 34, pp. 206-224a. 8. Fisher, R.A. (1959), Smoking, the cancer controversy. Some attempts to assess the evidence. Oliver and Boyd, London. 9. Doll, R., Hill, A.B. (1964), "Mortality in relation to smoking: 10 years' observations of British doctors", Br Med J, 1(5395): pp. 1399-1410 and 1(5396): pp.-1+6-0-1467. 10. Brownlee, K.A. (1965), "A review of Smoking and Health", J Amer Stat Assoc, Vol. 60, pp. 722-739.

References Page Two 11. Kahn, H.A. (1966), "The Dorn study of smoking and mortality among U.S. veterans: report on 8 years of observation", Nat'l Cancer Inst Monograph 19, W. Haenszel, ed., pp. 1-125. 12. Hammond, E.C. (1966), "Smoking in relation to the death rates of 1 million men and women", Nat'l Cancer Inst Monograph 19, W. Haenszel ed., pp. 12 0. 13. The Health Consequences of Smoking -1967, A Public Health Service Review. 14. Katz, L. (1967), "Statistical decision: theory versus application, exemplified in smoking-health controversy". Proc Int Stat Inst., Sydney, Australia. 15. Truett, J., Cornfield, J. Kannel, W. (1967), "A multi- variate analysis of the risk of coronary heart disease in Framingham", J Chron Dis, Vol. 20, pp. 511-524. 16. Seltzer, C.C. (1968), "An evaluation of the effect of smoking on coronary heart disease", J.Amer.Med. Assoc., Vol. 203, pp. 193-200. 17. Selzter, C.C. (1970), "The effect of cigarette smoking on coronary heart disease. Where do we stand now?", Arch Environ Health, Vol. 20, pp. 418-423. 18. Smokin and Health Now: a re ort of the Ro al Colle e of Physicians, London (1971). 19. Seltzer, C.C. (1972), "Critical appraisal of the Royal College of Physicians' report on smoking and health", The Lancet, pp. 243-248. 20. Seltzer, C.C. (1975), "Smoking and cardiovascular disease", Amer Heart J, pp. 125-126. 21. Burch, P.R.J. (1979), "Coronary heart disease: risk factors, age, and time", Amer Heart J, Vol. 97, No. 4, pp. 415-419. 22. Seltzer, C.C. (1979), with G.D. Friedman, A.B. Siegelaub, L.G. Dales. "Characteristics predictive of coronary heart disease in ex-smokers before they stopped smoking: comparison with persistent smokers and nonsmokers", J Chron Dis, Vol. 32, pp. 175-190.