Tobacco Documents Online

. were controlled... The data can be regarded as evidence against the constitutional hypothesis." Calculations are presented to note that observations made during the study suggest the possibility that data from respondents (those who answered the smoking questionnaire) may in fact underestimate the risk associated with smoking. The Surgeon General's 1964 Report had considered the possibility that differences between respondents and non-respondents to the questionnaire might have introduced a bias and had attempted to calculate a maximum estimate of that bias. Study of Men and Women in 25 States (This report is based on 3,764,571 person-years of experience and 43,221 deaths occurring among 1,003,229 subjects -- 440,558 men and 562,671 women -- between the ages of 35 and 84 f rom October 1, 1959, to February 15, 1960, when they enrolled in a prospective study and answered detailed questionnaires including questions on their smoking habits. Hammond (11).) "Death rates of both men and women were higher among subjects with a history of cigarette smoking than among those who never smoked regularly. "Death rates of current cigarette smokers increased with number of cigarettes smoked per day and degree of inhalation. "Death rates were higher among current cigarette smokers starting the habit at a young age than among those starting the habit later in life. Among both men and women, the difference between the death rates of cigarette smokers and non-smokers increased with age. "Among men, the death rates for ex-cigarette smokers were lower than for men currently smoking cigarettes when they enrolled in the study. Death rates of ex-cigarette smokers decreased with the length of time since they last smoked cigarettes." 11

"... Total death rates and death rates from most of the common diseases occurring in both sexes were higher in men than women, were higher in men who never smoked regularly than in women who never smoked regularly, and were far higher in men with a history of cigarette smoking than in women with a history of regular cigarette smoking. "The difference between the death rates of subjects with a history of cigarette smoking and subjects who never smoked regularly was far greater among men than women. Female cigarette smokers (as a group) have been far less exposed to cigarette smoke than male cigarette smokers of the same ages, as judged by number of cigarettes smoked per day, degree of inhalation, and the number of years they have smoked. Many female cigarette smokers smoke only a few cigarettes a day, do not inhale, and have been smoking for only a few years; their death rates are about the same as the death rates of women who never smoked regularly." Study of British Physicians: (The mortality of nearly profession in the United years. During the first of the women died. These 41,000 men and women in the medical Kingdom has been followed for 12 ten years 4,597 of the men and 366 deaths were analyzed in relation to smoking habits reported by doctors in reply to a questionnaire sent to them in 1951 --both sexes-- and again in 1957, men, and 1960, women. Doll and Hill (8,9).) "... An association with smoking is found, in differing degrees, in men for seven causes of death [which accounted for 39 percent of the death rataJ --namely, cancer of the lung, cancers of the upper respiratory 12

r and digestive tracts, chronic bronchitis, pulmonary tuberculosis, coronary disease without hypertension, peptic ulcer, and cirrhosis of the liver and alcoholism. No association is found with the remaining 61 percent of the death rate, and this includes such major causes as other forms of cancer, cerebrovascular accidents, hypertension, myocardial degeneration, suicide, and accidents. "In women, the few deaths at present available show an associa- tion only between smoking and cancer of the lung." "... If the excess deaths in smokers under the age of 65 years from (a) cancer of the lung, (b) chronic bronchitis a nd emphysema, and (c) coronary thrombosis without hypertension be taken as attributable to their cigarette smoking, then the total mortality from all causes at ages 45-64 years is increased thereby by approximately 50 percent." The report states: "One of the striking characteristics of British mortality in the last half-century has been the lack of improvement in the death rate of men in middle life. In cigarette smoking may lie one prominent cause." Study of Canadian Pensioners: (The purpose of the study was to investigate the relationships between residence, occupation, and smoking habits, and mortality from chronic diseases particularly lung cancer. It was initiated by a questionnaire which was sent to Canadian veteran pension recipients during the period September 1955, through June 1956. 13

Returns from 78,000 men, and 14,000 women, mostly widows, were analyzed. The men were mainly World War I and World War II veterans, but some Boer War and Korean War veterans, as well as some non-veteran pension recipients were included. The age of most of the men at the beginning of the study ranged from 30 to 90 years and the distribution was characterized by the ages of men eligible for se rvice in the two World Wars. For each respondent dying between July 1, 1956, and June 30, 1961, the cause of death was related to information on his questionnaire about age, history of smoking habits, residence and occupation. Among the respondents during the six years of follow-up there were 9,491 deaths of males, and 1,794 deaths of females which were analyzed (13).) - "Current cigarette smokers had a death rate for overall mortality 54 percent higher than that of non-smokers...Ex-cigarette smokers had a comparatively lower rate, which was still 36 percent above the rate for non-smokers...Men smoking combinations of cigarettes plus cigars and/or pipe also had elevated death rates for overall mortality, but these were not elevated to the same extent as those of men smoking only cigarettes." "The death rates for overall mortality of pipe smokers and cigar smokers were not appreciably different from those of non-smokers." "For cigarette smokers as compared to non-smokers., overall mortality ratios were elevated after five years of smoking at any time in their life and remained elevated as long as they continued to smoke cigarettes. " "Male current cigarette smokers who inhaled had a death rate for overall mortality 52 percent higher than that of those who did not inhale." "An urban/rural comparison was made between males of equivalent cigarette smoking habits and non-smokers. It was found that the death 14 I

t rate for overall mortality of urban dwellers (persons with a history of five years or more of city residence) was 12 percent higher than that for rural dwellers of comparable smoking habits." "Respondents were classified into occupational groups based on their history of occupation. No evidence was found in this study of clear-cut associations between cause of death and occupation. Further, occupation did not appear to modify the established association of cigarette smokers with death rates in excess of those of non-smokers." SOME GENERAL CONSIDERATIONS The problem of how best to measure the relationship between smoking and mortality has been discussed in the Surgeon General's 1964 Report as well as in some of the prospective study reports. As the amount of data available increases, the person-years of observations in the many population sub-groups that are worth examining increases so that stable rates may be computed and compared. A brief discussion of three measures of comparison available and their utility seems desirable as confusion frequently arises over these measures. (a) Mortality Ratios: Obtained by dividing the death rate for a classification of smokers by the death rate of a comparable group of non-smokers. (b) Differences in Mortality Rates: Obtained by subtracting from the death rate for smokers, the death rate of a comparable group of non-smokers. (c) Excess deaths: Obtained by subtracting from the number of deaths occurring in a group of smokers, the number of deaths w}ris h would have occurred if that group of smokers had experienced the same mortality rates as a comparable group of non-smokers. In the example which follows this has been reported as a percentage of all deaths in the appropriate age group. 15 i

Table 1 presents in summary form all three measures for 5 age groups of men from both the U.S. veterans study and Saaaond's study and for the same age groups of wosien from the latter study. The statistics were derived from the cited publications to make for reasonable comparability and may vary slightly from the figures combined in other ways. Also it should be noted that the age groups are not defined identically and the experience reported covers souewhat different tisye periods. The smoking group analyzed is "current cigarette smokers," i.e., those who were smoking at the time of enrollment into the study, and the comparison group is "never smoked regularly," i.e., those who had never been regular smokers of any form of tobacco. The number of deaths in each age-sex group is given to indicate the relative stability of the figures in that column. The data in the veterans study are largely concentrated in age groups 55-64 and 65-74. In Ha..ond's study, age group 35-44 is less stable than the succeeding groups both for men and for wowen. 16

Table 1 Comparison of Three Measures of Relationship between Cigarette Smoking and Overall Death Rates by Age and Sex as Derived from Two Major Prospective Studies (U, 13) * (a) Mortality Ratios -- Death Rate for Current Cigarette Smokers divided by Death Rate for those who Never Smoked Regularly (b) Difference in Death Rates -- Death Rate for Current Cigarette Smokers minus Death Rate for those who Never Smoked Regularly (c) Excess Deaths among Current Cigarette Smokers (i.e., additional deaths that occurred among current cigarette smokers per year above those which would have occurred if smokers had the same death rates as those who never smoked regularly). This is expressed as a percentage of all deaths occurring in that age-sex group. U.S. Veterans: Men Age 35-44 45-54 55-64 65-74 75-84 Total Deaths 383 366 13,840 17,550 1,932 Death Rates: Never Smoked 127 264 1,056 2,411 6,214 per 100,000 Regularly Death Rates: Current Ciga- 232 728 1,819 4,032 8,471 OD per 100,000 rette Smokers w (a) Mortality Ratio 1.83 2.76 1.72 1.67 1.36 ©~ (b) Difference in Death Rates 105 464 763 1,621 2,257 N per 100,000 (c) Excess Deaths as Percentage 33% 43% 21% 17% 8% of Total

Hammond • Men Total Deaths Death Rates: Never Smoked per 100,000 Regularly Death Rates: Current Ciga- per 100,000 rette Smokers (a) Mortality Ratio (b) Difference in Death Rates per 100,000 (c) Excess Deaths as Percentage of Total Hammond • Women Total Deaths Death Rates: Never Smoked per 100,000 Regularly Death Rates: Current Ciga- per 100,000 rette Smokers (a) Mortality Ratio (b) - Difference in Death Rates per 100,000 (c) Excess Deaths as Percentage of Total Table 1 Continued Age 35-44 45-54 55-64 65-74 75-84 631 5,297 8,427 8,125 3.968 210 406 1,202 3,168 7,863 397 925 2,202 4,788 9,674 1.89 2.28 1.83 1.51 1.23 187 519 1,000 1,620 1,811 33% 38% 25% 13% 4% 727 2,826 3,915 5,115 4,188 165 304 698 1,913 5,914 186 384 838 2,229 5,846 1.13 1.26 1.20 1.17 0.99 21 80 140 316 68 5% 9% 4% 2% -- * These figures are derived from the references. Five-year age groups were comtbined directly from the reported statistics without adjustment to any standard population. 9V0IC90e

9 measure of personal health significance, a means for the individual to estimate the added risk to which he is exposed. 3. Excess Deaths as a Percentage of Total Deaths: As with mortality ratios, this statistic appears to be highest in the age group 45-54 where it reaches 43 percent in one group of men and 38 percent in 1. Mortality Ratios: For men, these are at their highest in age group 45-54, diminishing in each subsequent decade. In both studies mortality ratios appear to be somewhat lower in the preceding decade 35-44. However, with the smaller numbers of cases available in that age group, it may be that selective factors contribute to the finding. For aosien the mortality ratios are much smaller than for ven, although the same pattern is suggested. In general, mortality ratios have been considered to reflect the degree to which a classification variable identifies or may account for variations in death rates. As such, it is a measure of relative risk which indicates the importance of that variable relative to uncontrolled variables -- an indicator of potential biological significance. 2. Differences in Mortality Rates: These increase consistently with increasing age in all three study groups, except for the oldest age group in women where there is practically no difference in the rates for smokers and non-smokers. Differences between smokers' rates and non- smokers' rates are much smaller for women than for men, as are the death rates themselves for men and women classified similarly with respect to smoking. This measure reflects the added probability of death in a one- year period for the smoker over that for the non-smoker. As such it is a 19

the other. Hammond's data by f ive-year age groups show the highest rate at ages 45-49, where it is 44 percent. Reviewing both study groups it appears tha t for men between the ages of 35 and 60 approximately one- third of all deaths that occur are excess deaths in the sense that they would not have occurred as early as they did if cigarette smokers had the same death rates as the non-smoking group. For wo.en, the percentage is auch lower, reaching a peak of 9 percent of all deaths in age group 45-54. It should be noted that this measure not only depends on the differences in death rates between the smokers and the non-smokers, but also on the proportion of smokers in the group. Thus, even with a large difference in rates between smokers and non-smokers, a popula- tion with very few smokers would have very few excess deaths. This measure is therefore an indicator of public health significance of the differences found since it measures the number of people affected and therefore the magnitude of the problem for society as a whole. Once the .agnitude of the excess is identified the problem becomes one of determining how much of the excess would not have occurred if it had not been for cigarette smoking and how much would have occurred anyhow. It should be noted that much of the excess has already been identified as belonging in the f irst category. Of the remainder, little of the excess has been clearly identified as belonging in the second category -- that is, not caused by smoking. With most of that remainder there is uncertainty as to the category in which it belongs. ~ © Ca O m 0 20