Document zZooROvMKnj18rdZwkB9q67

05$V2fc08 17:07 FAX ASBESTOS EXPOSURE, CIGARETTE SMOKING AND DEATH RATES* E, Cuyier Hammond.t Irving J. Selikoff,# and Herbert Seidmanf t Department ofEpidemiology md Statistics American Cancer Society . New York, New York 10017 ^Environmental Sciences Laboratory Mount Sinai School ofMedicine The City University ofNew York New York. New York 10029 Before 1966, there was evidence from a number of studies conducted in several different countries that people with a history of occupational exposure to asbestos dust are at risk of developing asbestosis and mesothelioma and at very high risk of lung cancer.14 There was also evidence that mesothelioma or lung cancer resulting from asbestos exposure does not usually appear until many years after initial exposure.5 Little more was known about the sometimes fatal effects of exposure to asbestos dust although there was some evidence that it increases the risk of cancer of the digestive tract.1 / This investigation was started in 1966 primarily to obtain information on the combined effects of cigarette smoking and exposure to asbestos dust in respect to death from lung cancer and chronic noninfectious pulmonary diseases; to obtain further evidence concerning cancer other than mesothelioma and lung cancer; and to obtain information on the degree to which occupational exposure to asbestos dust increases lota! death rates from all causes combined. Some of the early findings have been published.5'7 In any study of the effects of exposure to asbestos dust it is necessary to have information on two groups of people: an exposed group and a nonexposed group usually called the "control group." Obviously, the two groups should be as alike as possible except in respect to asbestos exposure. In mortality studies, the total , population is often used as the control group; and age-sex specific death rates as officially reported for the total population are compared with the age-sex specific death rates of asbestos workers. When this is done (and we have done it) it is only because of lack of availability of a more suitable control group. For this investigation, we could not use the total population as the only control group for the reason that we had to have information on the age-specific death rates of men with each of various types of smoking histories; and such information is not available for the total population. Therefore, we had to obtain a more suitable control group described later. The Exposed Group This investigation was made possible by the splendid cooperation of the Interna tional Association of Heat and Frost Insulators and Asbestos Workers that has about `Support to the Environmental Sciences Laboratory for this research has been available by a grant from the American Cancer Society, R-53. Additional support has been given by a grant to the Laboratory from the National Institute for Occupational Safety and Health, OH 00320. and by support to the Laboratory from the National Institute of Environmental Health Sciences, Center Grant ES00928. 473 0017-S923/79/Q330-0473 SI.75/0 1919 NYAS 05/15/2WB 17:07 FRX @012/030^ 474 Annals New York Academy of Sciences 120 locals in the United States and Canada. Until recently, most of the material handled by these insulation workers contained asbestos. Therefore, wo will refer to all of the members as asbestos workers. The union supplied us with their complete membership list and in 1966 we wrote to each member requesting him to complete a questionnaire containing a numtaer of questions including those on his smoking habits and his use (or nonuse) of protective masks. Date of birth and date of entry into the trade were ascertained from union records. All of these men have been traced through December 31, 1976 and copies of the death certificates of those who died have been obtained. Since cause of death recorded on a death certificate is not always accurate, we wrote to the doctor who signed the certificate requesting further details on diagnosis as well as to the hospitals in which treatment may have been given, requesting them to lend us histologic slides and/or x-ray films. Treating physicians, pathologists, hospitals were extremely cooperative in this matter. Some of the men on the 1966 membership list died before January 1, 1967 leaving 17,800 alive at that dale (Table 1). Altogether 2271 died in the ten-year period Table 1 Number of Men, Man-Years of Observation, Observed Deaths and Average age During Observation, Total, anp Up To ano More Than 20 Years from Onset of asbestos Exposure in a Cohort of asbestos Insulation Workers in the United States and Canada, JanuaRV i, 1967-Decembbr 3), 1976 Number of men Number of man-years Number of deaths Average age during observation Total 17,800 166,853 2,271 44.4 Asbestos Exposure* <20 Years 20+ Years 12,68? 89,46? 325 36.3 12,051 77,391 1,946 53,8 *As of January 1, 1967, 5117 men had reached the 20+ year point from onset of occupational exposure to asbestos dust. An additional 6934 men reached that point at some time between January 1,1967 and December 31, 1976, January 1, 1967---December 31, 1976; the number of man-years of observation totaled 166,853; and the average age of the men during the ten-year period was 44.4, Of the 17,800 men, 5117 bad entered the trade 20+ years prior to January 1, 1967; and 1573 of these died before January 1, 1977. Between January 1, 1967 and December 31, .1976, another 6934 men passed the 20-year point since entering the trade; and 373 of these died in the interval between the time they passed the 20-year point and December 31, 1976-Thus a total of 5117 + 6934 * 12,051 men had either reached the 20-year point prior to January 1, 1967 or reached that point before January I, 1977. The number of man-years of experience past the 20-year point totaled 77,391; the average age during observation was 53.8; and there were 1946 deaths. This report is confined to the 77,391 man-years of observation and the 1946 deaths just mentioned. The Control Group Starting on October 1,1959,468,688 men and 610,206 women were enrolled in a long term prospective epidemiological study by volunteer workers of the American Cancer Society in 1121 counties in 25 states/ All of them were over 30 years old at 05/15/2006 17:08 FAX Hammond et al.: Asbestos and Cigarette Smoking 475 that time and most of them were over 40. Upon enrollment, each subject answered a detailed questionnaire; and most of the survivors answered repeat questionnaires distributed in 1961, 1963, and 1965, During that time, death certificates were obtained oir these who died; and when cancer was mentioned on a death certificate, the doctor who signed the certificate was requested to supply additional information on the cause of death and the basis of the diagnosis. Follow-up was then discontinued for six years. Tracing of the subjects was resumed on October 1, 1971 and was continued through September 30, 1972. (We are still following subjects who were very old at that time). One of the major reasons we resumed the tracing was to obtain a suitable control group for the present investigation. Because of the extremely large number of deaths after 1965, it was not feasible for us to request doctors to supply us with additional information on cause of death. The selection of the subjects was such that persons at the bottom Of the social scale were almost entirely excluded (e g., impoverished migrant workers, illiterates, illegal immigrants, prison inmates, etc.). Persons in these categories are not members of the asbestos insulation workers union; but they are included in the mortality statistics of the United States, and their death rates are very high. On the other hand, unlike the asbestos workers, a largo proportion of the male subjects was employed in sedentary occupations involving little or no physical exertion; and lack of exercise is associated with increased risk of coronary heart disease"--the leading cause ofdeath among males in the United States, In contrast, men lacking a college education and with a history of occupational exposure to dust, fumes, gases, chemicals, or radiation are most likely to be in occupations involving physical exertion; and in this respect they are similar to members of the asbestos insulation workers union. Some such exposures increase the risk of cancer or pulmonary disease. Most of the members of the asbestos insulation workers union were white and virtually none of them were termers. For these reasons we selected as a control group for this investigation all of the male subjects who met the following specifications; white, not a termer, no more than a high school education, a history of occupational exposure to dust, fumes, vapors, gases, chemicals or radiation and alive as of January 1, 1967 and traced thereafter. There were 73,763 such subjects. They were classified according to their smoking histories and the age-specific death rates of each such class was computed. There was a problem. Death rates in the control group were known for the period January 1, 1967 through September 30, 1972 while members of the asbestos insulation workers union were traced during the period January 1, 1967-Decembcr 31, 1976. According to official mortality statistics, death rates of the general population of the United States changed somewhat during the second five-year period January 1, 1972-December 31, 1976 compared with the first, January I, 1967December 31, 1971, increasing for some diseases and declining for other diseases. Under the assumption that these changes probably also applied to the control group, the death rates of the control group were extrapolated to take this into account. The principal effect of the extrapolation was to increase the death rates from lung cancer and decrease the death rales from heart disease during the last five years as compared with the first five years. Causes of Death Because of the pressure to fill out a death certificate promptly so that the body may be moved, what is recorded on the certificate is not always the best available information on cause of death. For example, in the absence of the patient's physician, the certificate may be signed by a doctor who knows less about the case; or an autopsy 0.5/15/2008 17:08 FAX 476 Annals New York Academy of Sciences 014/03g.,g may indicate that the tentative diagnosis of cause of death was incorrect; or a review of the evidence may indicate for example, that the patient died of peritoneal mesothe lioma involving the pancreas rather than having died of primary carcinoma of the pancreas. Fortunately, we were able to obtain additional medical information on almost all of the asbestos workers who died; and in many instances histologic sections or x-ray films were loaned to m for further review. The first column of figures in Table 2 shows the 1946 deaths classified by underlying cause of death according to what was written on the death certificates, following as nearly as possible the coding rules used by the United States National Center for Health Statistics. The second column shows these same deaths classified according to the best available medical information, including, in many instances. Table 2 Deaths Occurring 20-f Years After Onset of Occupational Exposure to Asbestos Dust. Cause of Death Coded in Two Different Ways Underlying Cause of Death All Causes Cancer, all sites Lung Pleural mesothelioma Peritoneal mesothelioma Mesothelioma not specified above Larynx, buccal and pharynx Esophagus Kidney Colon-rectum Stomach / ' Prostate Bladcr Pancreas Other specified sites Primary site unknown Noninfeciious pulmonary diseases, total Asbestosis Cardiovascular diseases Other and unspecified causes Subtotal--all causes except cardiovascular diseases Based on Death Certificates Only (DC) 1,946 845 397 23 24 54 21 17 15 54 18 24 7 46 110 35 J 77 76 638 286 1,308 Based on Best Evidence Available (BE) 1,946 912 450 61 109 0 27 17 16 55 21 26 9 2! 83 17 204 160 566 264 1.380 Difference (BE)-(DC) +67 +53 + 38 +85 - 54 +6 0. +1 +1 +3 +2 +2 25 -- 27 -18 +27 +84 -72 -22 +72 review of histologic sections by pathologists with a special interest in mesothelioma and review of x-ray films by physicians with a special interest in asbestosis. In some instances, two or more pathologic conditions are present in such degrees that any one of them alone or the combined effects might have been responsible for death. Our procedure is to code up to three such conditions (i.c,, diseases or injuries). Following international custom, we designate one of these as the underlying cause of death; but this is sometimes a highly arbitrary decision. The third column of figures in Table 2 shows the difference between the first column, DC (death certificate), and the second column, BE (best evidence). There were major discrepancies in the comparative figures. According to death $5/15/2008 17:08 FAX Hammond et al.: Asbestos and Cigarette Smoking All certificates 845 deaths were due to cancer, while according to the best evidence available 912 deaths were due to cancer, a difference of 67 deaths. On the other hand, according to DC, 638 deaths were due to cardiovascular diseases (including stroke) while according to BE only 566 deaths were due to cardiovascular diseases, a difference of 72 in the opposite direction. Cancer of the pancreas accounted for 46 deaths (DC) or 21 deaths (BE), The best available medical information and review of histologic sections indicated that of the 46 deaths ascribed to primary cancer of the pancreas according to the death certificates, 15 were actually due to peritoneal mesothelioma, 6 were due to cancer of some other specified primary site and five to unknown primary site. One death attributed to myocardial infarction (DC) was probably due to primary cancer of the pancreas. Cancer of other specified sites accounted for 110 deaths (DC) or 83 (BE) a difference of 27, Cancer in which the primary site was unknown was assigned 35 deaths (DC) or 17 deaths (BE). Additional medical evidence (sometimes from an autopsy) accounted for this difference. Mesothelioma accounted for 101 deaths (DC) or 170 deaths (BE), a difference of 69. Most of the extra 69 deaths had been recorded on the death certificate as doe to cancer of some other primary site or "cancer, primary site unknown,'' The category labeled "noninfectious pulmonary diseases" includes chronic bronchitis, emphysema, pulmonary fibrosis, other nonspecific pulmonary diseases, and pneumoconiosis including asbestosis, This group of diseases caused 177 deaths (DC) or 204 deaths (BE) a difference of 27, The subcategory, asbestosis was termed the underlying cause of death for 76 (DC) or J160 (BE) of these deaths. Observed vs. Expected Number of Deaths Tables 3 and 4 show the 1946 observed deaths classified by cause of death in the two different ways previously discussed: from death certificates (DC) and according to the best evidence available (BE). Two different sets of expected numbers of deaths are shown: the expected number based on the age and smoking history specific death rates of men in the control group; and the expected number based upon the age-specific death rates of all white males in the United States as reported by the National Center for Health Statistics for the period 1967-1976, The age standardiza tion was by five-year age groups in each of two five-year intervals of time (1967-1971 and 1972-1976), Thus altogether four sets of comparisons between observed and expected numbers of deaths are shown. This was done because there may be differences of Opinion as to which set is preferable. In this analysis, in most instances, we prefer comparisons of the expected deaths based on the control group rates with the observed deaths classified according to the death certificates. Our reasons arc: the control group subjects were similar to the asbestos workers in respect to education and in respect to general type of work (i.e. work involving physical exercise and a history of exposure to dust etc.), smoking habits were taken into consideration, and although death certificates are not always accurate, when two groups are compared it is highly desirable that information on the causes of death come from the same source. However, the set of comparisons just mentioned leads to overestimation of the degree of association between exposure to asbestos and death from cancer of the pancreas while it underestimates the impact of mesothelioma and asbestosis. For these diseases, we prefer the comparisons where cause of death was classified according to the best evidence available (BE). Two indices of association are shown: the mortality difference which is the observed number of deaths minus the expected number of deaths, and the mortality 05/15/2008 17:08 FAX 478 Armais New York Academy of Sciences ii018/03g,,g ratio which is the observed number of deaths divided by the expected number of deaths. The general findings are shown in Table 3. For total deaths (disregarding cause of death) the mortality ratios were 1.70 (control group) and 1.41 (U.S. white males) and the mortality differences were +798.0 and +570.0 respectively. For cancer of all sites combined, the mortality ratios ranged from 3,05 (U.S. white males, DC) to 3.52 (control group, BE) and the mortality differences ranged from +567.9 to +653.0. Thus, according to every comparison, total death rates and death rates from cancer were elevated in the asbestos workers. The expected.number of deaths from noninfectious pulmonary diseases was higher for the control group (68.2) than for U.S. white males (53.8). This is not surprising since men in the control group had a history of occupational exposure to dust, fumes, vapors, etc. The mortality ratios ranged from 2.60 (control group, DC) to 3 79 (U.S. white males, BE) and the mortality difference ranged from +108.8 to +150.2. Asbestosis clearly accounted for these mortality differences. However, it seems likely that many of these deaths were due to the combined effects of asbestosis and damage to the lung parenchyma resulting from cigarette smoking. Evidence on this will be shown later in this paper. In contrast, the expected numbers of deaths from cardiovascular diseases were lower for the control group (660.1) than for U.S. white males (752.7), We do not know the reason for this, but it may have been due to the fact that men in the control group had a history of occupational exposures usually involving physical exercise (in contrast to clerical work). The mortality ratios ranged from .75 (U.S. white males, BE) to .97 (control group, DC) and the mortality differences ranged from --22.1 to -186,7. Considering all causes of dealh except cardiovascular disease, the mortality ratios ranged from 2.10 (U.S. white males, DC) to 2.83 (control group, BE). Table 4 shows the findings for cancer of several different sites. For lung cancer, the mortality ratios ranged from 4.24 (U.S. white males, DC) to 5,51 (control group, BE) and the mortality differences ranged from +303,3 to +368-3. Thus, no matter which of the various comparisons is considered to be the most reliable, it is clearly apparent that death rates from lung cander were greatly increased in this group of asbestos workers--enough so as to cause a significant decrease in their life expectan cy. From the standpoint of longevity, this was the most important effect of exposure to asbestos dust in a group of workers, the majority of whom had a history of cigarette smoking. Death rates from mesothelioma are not available for either the control group or U.S. white males; but they are presumably extremely low in these two groups, so the expected numbers of deaths from mesothelioma were probably less than 0J. Conse quently the mortality ratios must be extremely high. The mortality difference ranged from a trifle less than 101 to a trifle less than 170. The former figure is certainly an underestimate while the tatter figure, which is based upon histologic evidence as well as other evidence, is probably very close to correct. According to this evidence, 61 of the asbestos workers died of pleural mesothelioma and 109 died of peritoneal mesothelioma. Cancer originating in the larynx, pharynx or buccal cavity is more readily diagnosed than cancer of some other sites; and in a previous study, deaths from cancer of these sites were generally reported correctly on death certificates. Therefore, we were surprised that according to the death certificates only 21 Of the asbestos workers died Of cancer of these sites, while according to the best available evidence 27 died Of these cancers. The mortality ratios ranged from 1,60 (U S. while males, DC) to 3-60 (control group, BE). Two of the mortality differences are statistically significant at the level of p <.05; the other two are not. Table 3 Observed and Expected Number of Deaths Occurring 20+ Years After Onset of Occupational Exposure to Asbestos Dust Underlying Cause of Death All causes Cancer Cardiovascular Noninfectious Pulmonary (total) Asbesiosis All other causes Subtotal--all causes cxc. cardiovascular Observed Deaths From? Total -- 1946 DC 845 BE 912 DC 638 BE 566 DC m BE 204 DC 76 BE 160 DC 286 BE 264 DC 1308 BE 1380 Control Grcupf Expected Deaths Observed Minus Expected Ratio Observed/ Expected II48.Q + 798.0 1.70 259.0 + 586.0 +653.0 3.26 3.52 660.1 -'+22.1 -94.1 0.97 0.86 68 2 + 108,8 + 13S.S * +740 + 160,0 2.60 2.99 -- -- 160.8 + 125.2 +103.2 1,78 1,64 487,9 +820.1 +892.1 2,68 2.83 U.S. White Males Expected Deaths Observed Minus Expected Ratio Observed/ Expected 1376.0 + 570.0 1.41 277.1 4 567.9 +634.9 3.05 3.29 752.7 -114.7 -186.7 0.85 0.75 53.8 f 123*2 + 150.2 3.29 3.79 +76.0 + 160,0 _ -- 292.3 -6.3 --28.3 0.98 0.90 623J +684.7 +756.7 2.10 2.21 ""Control group and U.S. death rates rot available but these have been rare causes of death in Ihe general population. tLike men in American Cancer Socieiy prospective study (see text). tDC means cause of death according to death certificate information. BE means cause of death according to best evidence available. Ham m ond et a t: Asbestos and C igarette Sm oking o(A hoo0>0 oCO Kyy1 Ta&le 4 Observed and Expected Number of Cancer Deaths Occurring 20-t- Years After Onset of Occupational Exposure to Asbestos Dust Underlying Cause or Death (Cancers) Observed Deaths Front:) Total Control Groapf Expected Deaths. Observed Minus Expected Ratio Observed Expected U.S. White Males Expected Deaths Observed Minus Expected Ratio Observed Expected Lung Mesothelioma Larynx, buccal and pharynx Esophagus Kidney Colon-rectum Stomach Prostate Bladder Pancreas ' Other and unspecified cancers DC 39? 81.7 BE 450 DC 101 * BE 17C DC 21 7.5 BE 27 PC 17 ti? IT 5,1 DC 15 8.5 BE 16 s- DC 54 30.5 BE 55 DC 13 12,5 BE 21 DC 24 1$3 BE 26 DC 7 6.7 BE 9 DC 46 16.0 BE 21 DC 145 72.1 BE ICO 315,3 3683 hm.0 170.0 133 193 11.9 11.9 6.5 7.5 23.5 24.5 5.5 8.5 5.7 7.7 03 23 30.0 5.0 72.9 27.9 4.86 531 -- _ 2.30 3.60 3.14 334 1.76 1,88 1.77 1.81 1.44 1.68 1.31 1.42 1.04 134 2,87, 131 201 i-39 93.7 * 13.3 6,5 7.0 34.1 12.7 19.7 8.6 15.5 663 3033 3563 101.0 170,0 7.9 33.9 103 10.5 8.0 9.0 19.9 20.9 5.3 8.3 43 63 -1.6 0.4 30.5 5.5 78.7 33.7 4,24 4.80 --V- 1-60 2.05 2.64 2.64 2.15 230 139 1.62 1.42 1.65 1.21 132 0.82 1.05 2.96 1.35 2,19 LSI Control group and ILS.draUh rates not available but these have bees rare causes of death in the general population. tlifce mm in American Cancer Society prospective study (see text). $DC means cause of death according to death certificate information BE means cause of death according to tel evidence available. 480 Annals New York Academy o f Sciences CO *DXT S 2 XOCO. gh a 05/15/2008 17:08 FAX Hammond el ai: Asbestos and Cigarette Smoking 48 J For dancer of the esophagus, the mortality ratios ranged from 2.64 to 3.34. The mortality differences ranged from +10.5 to + 11.9 and are statistically significant. Without much doubt, exposure to asbestos dust increased the risk of this disease in a group of men, the great majority of whom smoked cigarettes, cigars or pipes. Coion-rectum cancer was in third place (after lung cancer and mesothelioma) as a cause of cancer death among the asbestos workers. The mortality ratios ranged From 1.59 to 1.81. The mortality differences ranged from 419-9 to +24,5 and are statistically significant (p <.05). For kidney cancer, the mortality ratios ranged from 1.76 (control group, DC) to 2.30 (U.S. white males, BE) and the mortality differences ranged from +6.5 to +9.0. Two of the mortality differences are statistically significant; the other two are not. Death rates from stomach cancer have been declining rapidly in the United States and only 18 (DC) or 21 (BE) of the asbestos workers died of this disease. The mortality ratios ranged from 1.42 to 1.68. The mortality differences ranged from +5.3 to + 8.5 and are not statistically significant (p > .05). Mortality ratios for cancer of the prostate ranged from 1.21 (U-S- white males, DC) to 1.42 (control group, BE), The mortality differences are not statistically significant. Very few of the asbestos workers died of bladder cancer and there is no indication that exposure to asbestos dust increases the risk of this disease. The figures for cancer of the pancreas are interesting because so many of the deaths attributed to this disease on death certificates were actually due to mesothelio ma. If the death certificate were to be believed, then it would appear that exposure to asbestos dust greatly increases the risk'of cancer of the pancreas. Better evidence suggests that death rates from cancer of the pancreas may be slightly elevated in asbestos workers, but this is far from certain. For cancer of ail other sites combined (including cancer of unknown primary sites) the mortality ratios ranged from 1.39 (control group, BE) to 2.19 (U.$, white males, DC). The mortality differences ranged from +27.9 to +78.7 and are statistically significant. Smoking Habits Table 5 shows the age distribution of the asbestos workers as of January J, 1967 and the smoking habits of those who answered the questionnaire in late 1966. It is confined to the 12,051 men who by January 1, 1967 had passed the 20-ycar point since entering the trade or who passed the point before January 1, 1977. Of the 8220 men who answered the questionnaire, 891 (11%) said that they had never smoked regularly; 488 (6%) Had a history of pipe or cigar smoking but never smoked cigarettes regularly and; 6841 (83%) were either current or ex-cigarcuc smokers. Some of the cigarette smokers also smoked pipes or cigars or had done so in the past. Our first question related to the degree to which cigarette smoking increased the death rates of the asbestos workers.5 This is shown in Tables 6A and 6B Tor all causes of death and for lung cancer identified as such from the best available medical evidence (BE) (the preferred classification here because all comparisons are within the asbestos workers group). The asbestos workers were first divided into five groups according to smoking habits: 1) smoked 20+ cigarettes a day; 2) smoked cigarettes regularly but < 20 a day; 3) history of regular pipe or cigar smoking but never smoked cigarettes regularly; 4) never smoked regularly; and 5) smoking habits unknown. The 20+ a day and the < 20 a day cigarette smokers are each subdivided into four groups: "current" meaning that they were smoking cigarettes regularly in 1966; ex-cigarette 0018/03^^0 Co> n X o-bo. Table 5 Number of Studv Group Men who Attained at Least 20 Years From Onset of Occupational Exposure to Asbestos Dust at Some Time January |, 1 967-Decemser 31, 1976, by Age January I, 1967, and by Smoking History as of 1966 Age as of Jan, 1, 1967 Total 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 30-84 85+ Total in Age Group 3 2,051 339 1,662 2,610 2,151 1,526 J,250 960 693 413 255 111 52 29 Cigarettes* 6,841 174 1,423 1,336 923 792 558 388 203 108 60 19 13 Current Cigarettes 20+/day 3,708 105 531 863 818 487 398 248 132 75 28 17 4 2 History of Smoking { Number of Men) Current Cigarettes <20/day Ex-Cigarettes 20+/day Ex-CigareUes <20/day Never Smoked Regularly 764 1,823 378 891 29 100 ,, 161 *114 94 74 70 63 27 15 12 4 3 31 160 313 329 265 253 193 134 74 42 18 5 6 9 43 65 58 49 44 33 35 . 16 15 6 4 I 27 142 198 143 !C7 84 65 50 33 29 5 3 5 Includes the cigarette smokers who did not specify the number smoked per day. Pipe/Cigar Only 488 6 66 84 69 61 46 40 44 31 29 5 5 2 Unknown History 3,831 132 610 905 603 435 328 297 211 146 89 41 25 9 Annals New York Academy o f Sciences fVoO, 04 as, 05/15/2006 IT:OS FAX Hammond el a!.: Asbestos and Cigarette Smoking 483 Table 6A Observed and Expected Number of Deaths Occurring 20+ Years After Onset of Occupational Exposure to asbestos Dust by Smoking History. Expected Deaths are Basso on Age-Specific Death Rates of All of the asbestos workers Smoking History Total* 20+ cigareites/day Current Ex, <5 yrs. Ex, 5-9 yrs. Ex, 10+ yrs. <20 cigarettes/day Current Ex, <5 yrs. Ex, 5-9 yrs. Ex, 10+ yrs. Pipe/cigar only Never smoked regularly Unknown All Causes of Death Observed Deaths Expected Deaths t946 1946.0 565 481,3 166 137.5 61 69,6 112 06.5 143 127.6 22 20.0 10 11.5 31 43.6 86 112.5 94 158.2 614 587.6 Lung Cancer (BE)t Observed Deaths Expected Deaths 450 450,0 171 116.8 54 33.3 10 17.1 15 31.4 38 29.5 5 4.8 1 2.5 5 9.8 3 24.2 5 35.5 06 132.6 0O21/O3 g^g i ; I i Table 6B mortality Ratios of observed and Expected number op Deaths Presented in Table 6A Relative to Total Men and to the Mortality Ratios for Men who Never Smoked Regularly (NSR) Smoking History Total* 20+ cigarettes/day Current Ex, <5 yrs. Ex, 5--9 yrs. Ex, 10+ yrs. <20 cigarettes/day Current Ex, <5 yrs. Ex, 5-9 yrs. Ex, 10 yrs. Pipe/cigar only Never smoked regularly Unknown All Causes of Death Ratio Observed / Expected Ratio to NSR 1.00 1.68 1.17 1.21 0.88 0.82 1.12 1.10 0.87 0.71 0,76 0.59 1.04 1.98 2.03 1.48 1.38 1.89 1.85 1,46 1.20 1.29 i.QO 1.76 Lung Cancer (BE)+ Ratio Observed/ Expected Ratio to ' NSR 1.00 7.10 1.46 1.62 0.58 0.48 1.29 1.04 0.40 0.51 0.12 0.14 1,03 10.40 11.51 4.15 3.39 9.15 7.40 2,34 3.62 0.38 I.O0 7,28 *tn addition to the smoking groups shown, the total includes the cigarette smokers who did not specify the number smoked per day, or the length of time stopped if an ex-smoker, f Died of lung cancer according to the best evidence available. 05/15/2008 17:08 FAX 484 Annals New York Academy of Sciences 0 022/03 jj,D smokers who had given up the habit for <5 years; ex-cigarette smokers who had given up the habit for 5 to 9 years; and ex-cigarette smokers who had given up the habit for 10+ years. TaBLE 6A shows the observed and expected number of deaths in each smoking category. The expected number for each particular category was calculated by applying the age-specific death rates of ail the asbestos workers (regardless of smoking history) to the age-specific man-years of exposure to risk for that particular smoking category. Table 6B, which is based upon Table 6A, shows two different ratios: the mortality ratio calculated by dividing the observed number of deaths by the expected number Of deaths (by definition, the mortality ratio for all of the subjects is 1.00), and the ratio relative to never smoked regularly, NSR, which is the mortality ratio for each smoking category divided by the mortality ratio for men who never smoked regularly (by definition this ratio for men who never smoked regularly is 1.00). The latter ratio is essentially the ratio which would have been obtained if men who had never smoked regularly had been used as a control group; but this method of calculation provides somewhat more stable figures. The pattern of relationships between smoking habits and death rates was found to be essentially the same for asbestos workers as for other groups of men as previously reported by numerous different Investigations. The following applies to deaths from ail causes combined; and the mortality ratios relative to NSR (never smoked regularly) (see Table 6B). The mortality ratios were 1,98 for men who currently smoked 20+ cigarettes a day, 1.8? for those who currently smoked <20 cigarettes a day, 1.29 for pipe and cigar smoking, and 1.00 for men who never smoked regularly. The mortality ratio for ex-cigarette smokers who previously smoked 20+ cigarettes a day and had given up the habit <5 years was 2.03. This is higher than the mortality ratio for men who currently smoked 20 + cigarettes a day. It is a usual finding and is almost certainly due to the fact that men with incurable lung cancer or other advanced disease oftettgive up smoking a few.months or longer before they die. The 20+ a day cigarette smokers who had given up the habit for 5 to 9 years had a mortality ratio of 1.48 and those who Jjad quit for 10+ years had a mortality ratio of 1.38, so their death rates were lower than the death rates of men who currently smoked 20+ cigarettes a day. The mortality ratios for ex-cigarette smokers who had previously smoked <20 cigarettes a day were; 1.85 (quit <5 years), 1.46 (quit 5 to 9 years), and 1.20 (quit 10+ years) as contrasted with a mortality ratio of 1.89 for men who currently smoked <20 cigarettes a day. For lung cancer, the pattern of relationship was about the same as described above, but the mortality ratios of cigarette smokers Were far higher. For example, the mortality ratio was 10.40 for men who currently smoked 20+ cigarettes a day and 9.15 for men who currently smoked <20 cigarettes a day. ' There were too few deaths from each of various other diseases to present such a detailed analysis in relation to smoking habits. Some of the findings are briefly described below using the ratio Observed/Expected. Only five men who never smoked regularly died of noninfcctious pulmonary diseases whereas I7.J deaths were expected in this group (ratio .29). These five died of asbestosis. Thus, for the subcategory asbestosis, there were five deaths among men who never smoked regularly whereas 13,3 were expected (ratio .38). Since all of the Other men were smokers (mainly cigarette smokers) it follows that death rates from asbestosis were far higher in cigarette smokers than in nonsmokers. The asbestosis death rates of men who currently smoked 20+ cigarettes a day was 2.8 times as high as the asbestosis death rate of men who never smoked regularly. It is clear that cigarette smoking greatly increases the risk of an asbestos worker dying 05/15/2008 17:08 FAX Hammond et al.: Asbestos and Cigarette Smoking 485 from asbestosis or asbcstosis combined with pulmonary fibrosis and emphysema resulting from cigarette smoking. None of the men who had never smoked regularly died of cancer of the esophagus, larynx, pharynx or buccal cavity. The expected number of deaths from these cancers was 3.6 for men who never smoked regularly. This suggests that in the absence of exposure to tobacco, exposure to asbestos dust may have iiulc or no influence on death rates from such cancers. For pleural mesothelioma, there were three observed deaths vs. 4.8 expected deaths among men who never smoked regularly (ratio ,63); and five observed deaths vs. 3.2 expected deaths for pipe/cigar smokers (ratio 1.59). Thus, for men who never smoked cigarettes regularly there were eight observed deaths and eight expected deaths from this disease; so it appears that cigarette smoking has little or no effect upon death rates from pleural mesothelioma. Oth rr Comparisons In Table 7A, the observed number of lung cancer deaths among asbestos workers in each of several different smoking categories is contrasted with the number expected if their age-specific lung cancer death rates had been the same as those for control group subjects who never smoked regularly. The observed numbers are based upon death certificate information. The tabic is confined to asbestos workers whose smoking habits were known. ; Altogether, according lo the death 'certificates, 276 lung cancer deaths occurred among the asbestos workers whose smoking habits were known, whereas only 6.0 were expected had their age-specific lung cancer death rates been the same as those for control group subjects who never smoked regularly. Thus, the mortality difference was 270,0 and the mortality ratio was 46.23. The mortality ratios for various smoking categories were 87.36 for asbestos workers who currently smoked 20+ cigarettes a day; 50,82 for those who currently smoked <20 cigarettes a day; 36.56 for ex-cigarette smokers; 7,02 for pipe/cigar smokers; and 5-33 for asbestos workers who never smoked regularly. Taken at face value it would appear that the risk of dying of lung cancer is about five times as great for a nonsmoker occupationally exposed to asbestos dust as for a nonsmoker without such exposure. However, because of small numbers the ratio is subject to considerable statistical sampling variation. By chance, the observed number might have been as low as one or two or as high as about nine. Therefore, all we can say is that the actual mortality ratio is probably > 1.00 but could be as high as 12.00 or even a little higher. In any event, the mortality ratio for men who never smoked regularly was yery small as compared with the mortality ratios of the cigarette smokers. The mortality differences shown in Table 7B are copied directly from Table 7A. Altogether, 270.0 more of the asbestos workers died of lung cancer than would have died had their lung cancer death rates been the same as for nOnsmokers not exposed to asbestos dust. These may be called "extra" lung cancer deaths. Of these "extra" deaths, 55.7% occurred in asbestos workers who currently smoked 20+ cigarettes a day; 11.3% in those who currently smoked <20 cigarettes a day; 29.9% in ex-cigarette smokers; 1.3% in pipe/cigar smokers; and 1.2% in asbestos workers who never smoked regularly. These figures are shown to give some idea of the degree to which the lung cancer death rate of a group of asbestos workers depends upon their distribution in regard to smoking habits. . O5/15/2608 1?:09 FAX 486 Annals New York Academy of Sciences 0 024/03^^ Table 7A Observed number of Lunq Cancer Deaths Occurring 20+ Years After Onset of Occupational Exposure to Asbestos dust by Smoking History Compared with Expected Number Based on Lung Cancer Death Rates of Control Group Subjects who Neves Smoked Regularly Smoking History Total (with smoking habits known) History of cigarettes? Current 20+ cigarettes/day Current <20dgarcttes/day Ex-cigarettes Pipe/cigar only Never smoked regularly Observed 276 268 152 31 83 4 4 Lung Cancer (DC)* Expected 6.0 4.7 1.7 0.6 2.3 0.6 0.7 Mortality Differences +270.0 + 263.3 + 130.3 + 30.4 + 80.7 + 3,4 +3.3 Ratio Observed/ Expected 46.23 57.63 87.36 50-82 36.56 7.02 5.33 Table 7B Distribution op Lunc Cancer Mortality Differences Presented in Table 7a by Smoking History Smoking History Total (with smoking habits known) History of cigarettes? Current 20+ cigarettes/d*y Current <20 cigarettes/day Ex-cigarettes Pipe/cigar only Never smoked regularly Lung Cancer (DC)* Mortality Differences Percent of Total +270.0 +263.3 , +150.3 / +30.4 + 80.7 + 3.4 +3.3 . 100.0 07.5 55,7 1U 29.9 1.3 1.2 *Dicd of lung cancer according to death ceriHieate information. tIncludes the current cigarette smoking asbestos workers who did not specify the number smoked per day. Table 8 differs from all of the other tables in that it shows lung cancer death rates per 100,000 man-years standardised for age on the age distribution of alt of the asbestos workers in this study. The lung cancer deaths were so reported on death certificates. Just four groups are shown identified by occupational exposure to asbestos dust, (yes or no) and history of cigarette smoking (yes or no). The word "no" under the "history of cigarette smoking?" means never smoked regularly (pipe and/or cigar only smokers being excluded from this table). The mortality differences shown here were calculated by subtracting the death rate of the "no, no" group from the death rate of each of the four groups. The mortality ratios were calculated by dividing the death rate of each group by the death rate of the "no, no" group. The mortality ratios are LOO for "no, no," (asbestos no; cigarette smoking, no); 5.17 for "yes, no" (asbestos, yes; cigarette smoking, no); 10,85 for "no, yes" (asbestos. s 05/15/2006 17:03 FAX Hammond et at.: Asbestos and Cigarette Smoking 487 no; cigarette smoking, yes); and 53.24 for "`yes, yes" (asbestos, yes; cigarette smoking, yes). Now, suppose that occupational exposure to asbestos dust and cigarette smoking acted independently in respect to the production of lung cancer. In that event, tho lung cancer death rate of asbestos workers with a history of cigarette smoking should be very close to the sum of the following three numbers; 11.3 (the rate for the "no, no" group), 47.1 (the mortality difference for the "yes, no" group), and 111.3 (the mortality difference for the "no, yes" group). The sum comes to 169.7 lung cancer deaths per 100,000 man-years which is a reasonable estimate of what the lung cancer death rate of the asbestos workers with a history of cigarette smoking would have been if there had been no synergistic effect of the combined exposure. In contrast, the observed lung cancer death rate of the "yes, yes" group was 601.6 per 100,000 man-years. The difference, (601.6 ~ 169.7) 431.9 lung cancer deaths per 100,000 man-years, was presumably due to a synergistic effect in men with both of the two types of exposure (asbestos dust and cigarette smoking). These particular figures apply to the particular group of asbestos workers included in this study. The net synergistic effect would not have been the same if their smoking Tabu; g Age-Standardized lung Cancer Death Rates* tor Cigarette Smoking and/or Occupational Exposure to Asbestos Dust Compared with no Smoking and no Occupational exposure to Asbestos Dust Group Control Asbestos workers Control Asbestos workers Exposure to Asbestos? No Yes No Yes History Cigarette Smoking? No No Yes Yes Death Rate 11J . 58.4 122.6 601.6 Mortality Difference 0.0 +47.1 + 111.3 + 590.3 Mortality Ratio 1,00 5.17 10-85 53,24 Rate per 100,000 man-years standardized for age on the distribution of the man-years of all the asbestos workers. Number of lung cancer deaths based on death certificate information, habits had been different; and it probably would not have been the same ff their lapsed times from first exposure to asbestos dust had been different or if the amount of asbestos dust they inhaled had been different. Comments Underlying cause of death cannot always be ascertained with certainty; when cancer is clearly the cause of death, the primary site of the disease cannot always be ascertained with accuracy; and even when histologic sections are available, different pathologists may disagree on the histologic type of a cancer. These difficulties arc troublesome from the standpoint of scientific accuracy as to details, but they make little difference to the workers and their families. For example, both peritoneal mesothelioma and cancer of the pancreas are painful diseases and the degree of pain is not dependent upon ascertaining which of these caused the death of the patient. If inadequate information results in error as to which of these two cancers caused death. 05/15/2008 17:08 0 028/03 488 Annals New York Academy of Sciences then it only means that the degree of association between death rates and exposure to asbestos dust is overestimated for one of the two diseases and correspondingly underestimated for the other. It is true that some diseases are mere painful than others and some are more expensive than others. But what is of most importance to the worker and most important to society is whether a particular type of occupational exposure leads to premature death or often leads to long term disability. Occupational exposure to asbestos dust results in both of these. This is sufficient reason to take strong preventive measures regardless of further details which are of scientific interest or which can have an influence on the outcome of a law suit. Evidence from this study indicates a strong synergistic effect between two types of exposure (asbestos dust and cigarette smoking) in respect to risk of lung cancer. Evidence from this study indicates that among asbestos workers, ex-cigarette smokers have substantially lower death rates than cigarette smokers who do not give up the habit. This should be brought forcefully to the attention of present asbestos workers. A young person who is so strongly addicted to cigarette smoking that he cannot break the habit or is unwilling to do so would be particularly well advised not to enter a trade involving exposure to asbestos dust. Cigarette smoking increases the risk of death from asbestosis, providing another reason why asbestos workers should avoid that habit. The major specific fata! effects of occupational exposure to asbestos dust are increased death rates from lung cancer, mesothelioma and asbestosis. The evidence indicates that occupational exposure to asbestos dust increases death rates from cancer of the colon-rectum, and cancer of the esophagus. It probably increases death rates from cancer of the kidney and from another set of cancers taken as a group (i.e., larynx, pharynx and buccal cavity). Aside from this, the evidence suggests that death rates from cancer of several other sites may be somewhat increased by occupational exposure to asbestos dust. Because of small numbers, we cannot identify which, if any, other specific sites are involved. As will be published in the near future by another group of investigators, asbestos bodies have.been found in many different organs of asbestos workers who died and whose lungs were found to contain a very large number of asbestos bodies.10 It must be emphasized that only one group of asbestos workers is included in this investigation, namely members of the insulation workers union. We have indicated the degree of confidence we have in various specific findings, and in many instances we have given more than one estimate. For example, we said that "the mortality ratios for colon-rcctum cancer ranged from 1.59 to 1.81," the mortality ratios having been calculated from each of four different sets of figures. In this particular case, a(l four sets of comparisons led us to the conclusion that death rates from colon-rectum cancer are increased by exposure to asbestos dust; but it would be folly to suppose that we have precisely determined the degree of association even in this group of asbestos workers. Aside from this, the degree of association between exposure to asbestos dust and death rates from all causes combined as well as death rates from lung cancer vary with smoking history, lapsed time from onset of exposure to asbestos dust,11 presumably with amount of asbestos dust inhaled, and probably with other factors as well. It seems very unlikely that any two identifiable groups of people with occupational exposure to asbestos dust are exactly alike in respect to all of the factors just mentioned. Therefore, wc would not expect the exact figures reported here to be duplicated in a study of any other group of asbestos workers. We would expect the findings to be qualitatively the same but not qualitatively identical. 05/15/20GB 17:10 FAX Hammond et aL\ Asbestos and Cigarette Smoking 489 Acknowledgments This research has been part of the Environmental Cancer Research Project of the American Cancer Society and the Mount Sinai School of Medicine, of which Dr. Hammond and Dr, SdikofT are Co-Directors. Many individuals assisted us in this research, and we are grateful to them. The leaders and members of the International Association of Heat and Frost Insulators and Asbestos Workers provided every help; Mr. Andrew T. Haas, International President, Mr. William 0. Bernard, General Secretary, Mr. Roy J. Steinfurth, coordinator of the Insulators Health Hazard Program and the officials of the local unions throughout the United States and Canada made it possible for us to maintain observation of each of the 17,800 men over the decade of study. They were an integral and essential pan of the research: We.wish to thank David Newman, Ashley Bodden, Edwin Silverberg and Marie Corbo of the Department of Epidemiology and Statistics of the American Cancer Society for their assistance! processing the data of this study. We also thank again the 68,000 volunteer reseWhers who helped make feasible the American Cancer Society prospective study which provided the control group in this report. Drs, J, Churg, V. Suzuki and M, Kannerstein, of the Mount Sinai Pathology Unit, reviewed the histological material; their advice and judgement constituted the skilled basis for pathological categorization and we appreciate their help. At the Environmen tal Sciences Laboratory, Janet S, Kaffcnburgh and Frances Perez diligently helped maintain observation of the insulation wdrkers cohort and collected and maintained the large number of specimens and x-rays that were needed. Our dedicated supporting staff at Mount Sinai was available throughout the investigation: Selma Annenberg, Richard Ashley, Doris Fleisher, Rupert Fuller, Judy Mariner, Maria Martinez, Diane Monahan, Vera Reitman, Julia Roberts, Albert Rodriguez and Sidney Sibel. We are also deeply appreciative of the generous help given to us by pathologists, clinicians, hospitals, coroners, state health departments and other medical facilities In all parts of the United States and Canada, providing detailed information, records, slides, blocks, roentgenograms, death certificates. While it is not possible to individu ally acknowledge this help here, each has our thanks and warm gratitude. References 1. Dou., R. 1955. Mortality from lung cancer in asbestos workers. Brit, J. Indust. Med. 12: 81-86. 2. Wagner, J. C-, C, A. Sleggs &. P. Marchand. I960. Diffuse pleural mesothelioma and asbestos exposure in the North Western Cape Province. Brit. J. Indust. Med. 17: 260 271. 3. Selikoff, 1. J,, j. Churg & E. C. Hammond. 1964. Asbestos exposure and neoplasia. J. Am. Med, Asso, 188: 22-26. 4. Selmcoff, 1. J.. J,, Churg & E. C. Hammond. 1965. Relation between exposure to asbestos and mesothelioma. New Eng. J. Med. 272? 560-565. 5- Seukgft, 1. I., R. A. Bader, M, E. Badf.r, J. Churg & E. C. Hammond. 1967, Editorial: Asbestosis and neoplasia. Amer. J, Med, 42:487-496. 6. Selikoff, 1. J., E. C. Hammond A H. Seidman. 1973. Cancer risk of insulation workers in the United States. /: Biological Effects of Asbestos, P. Bogovski el at., Eds.: 209 216, IARCSc. Pub. No 8, Lyons, France. 7. Hammond. E. C. & I. J. Selikoff, 1973. Relation of cigarette smoking to risk of death of asbestos-associated disease among insulation workers in the United States. In: Biological OS/15/2O0B 17:10 FAX 0 028/03 490 Annals New York Academy of Sciences Effects of Asbestos. P, Bogovskt et aL, Eds.: 312-317. IaRC Sci. Pub. Mo. g, Lyons, France8. Hammond, E. C. 1966. Smoking in relation to the death rates of one million men and women. In: Epidemiological Study of Cancer and Other Chronic Diseases, Monograph 19:127-204. National Cancer Institute. 9. Sbukopp, I. 3-, E. C. HAMMOND & j. Chukg. 1968, Asbestos exposure, smoking and neoplasia. J. Am. Med. Asso, 204;106-112. 10. Auerbach, 0-, A, S. Conston, L. Garfjnkei., V. R. Parks. H. D. Kasiow & E. C. Hammond. Presence oFasbestos bodies in organs other than the lungs. To be published. 11. Selikofp, 1. J,. E. C. Hammond & H. Shipman. 1979, Mortality experience of insulation workers in the United States and Canada, 1943-1976. Ann. N.Y. Acad. Sci. This volume. ji