Document JJnJRJM1mDyLemKnX7zKw7mOe

Exhibit 3 JPATTOfsTAt -v.kunmeNTAL health Montreal, Canada Mortality in the Chrysotile Asbestos Mines and Mills of Quebec ]. Corbett McDonald. MD; Alison D. McDonald, MD: Graham IF. Gibbs. MSc; Jack Siemiatvcki; and Charles E. Rassiter, MA, Montreal 30^2 677 O *3 10 % 3r 7c 70 to 0 i Of 11,788 persons born between 1891 and 1920 employed in the Quebec asbestos mining indus try, 88.4% were traced. Of these 2,457 (23.6%) had died. Exposure indexes for each worker were calculated from job dust levels and duration of employment. The overall mortality was lower than expected for the population of Quebec but in the highest dust category, comprising 5% of the cohort, the age-standardized rate was 20% higher than in the other groups. Respiratory, cardiovas cular, and malignant disease in equal proportions accounted for the excess. There were 101 deaths from respiratory cancer including three from ma lignant mesothelioma, an estimated excess of about 15 deaths. The difference in rates for re spiratory cancer between those maximally and minimally exposed was fivefold and, though per haps exaggerated, was apparently determined by accumulated dust exposure and duration of em ployment. The REMARKABLE qualities of the as bestos group of fibrous minerals have been recognized since antiquity, but mining and milling on an industrial scale began only at the end of the 19th century. In the Eastern Townships region of Quebec, deposits of chrysotile asbestos in serpentine rock were Submitted for publication Aug 17, 1970; accepted Nov 10. From the Department of Epidemiology and Health, McGill University. Montreal. Mr. Rossiter Is presently with the Medical Research Council Pneumoconiosis Unit, Penarth, South Wales. Reprint requests to 3775 TJniveisilv St, Montreal 112 tDr. J. C. McDonald). noted in the 1847 Canadian Geological Sur vey. The first mine was opened at Thetford in 1878, and within SO years the region was producing most of the world's asbestos.'The proportion fell as Russian, South African, and Italian mines came into operation, but Quebec still produces about 40% of the world's supply, now estimated at about 4 million tons a year.1 There are two main mining areas, one at Thetford Mines and neighboring towns of Black Lake and Broughton, and the other at Asbestos. The Thetford area was devel oped by many different companies, but with amalgamation the number has now been re duced to six. At Asbestos, the mining has been carried out since 1882 by one large company which also operates a small fac tory in the town for the manufacture of mixed asbestos products. There is a small mine owned by another company a few miles away. Concern for the health effects of asbestos has paralleled growth in its production, and the main available evidence was reviewed fully by Wright3 in 1969. The first cases of diffuse pulmonary fibrosis after prolonged exposure were noted by Murray in 1907. and by 1930 asbestosis was recognized as an important occupational hazard. Controls in troduced since then have reduced consider ably the dust concentration to which workers in the mines, mills, and primary manufac- Arch Environ Health--Vot SJlTZIT 4 SXYTSOSsGST 2^',. EliJ.T3, U2RABS ASARCO ALV 0005732 67S MORTALITY AND CHRYS0T1LE MINES--McDONALD ET AL turing industries are exposed. Though the prevention of asbestosis is far from complete in these industries, it is still generally be lieved that this can be achieved by more strict environmental control. Another aspect of the problem appeared in 1947 when Merewether3 showed a link between asbes tos exposure and lung cancer. The potential extent of this hazard was increased by re ports of Wagner and others4 during the last ten years that malignant tumors of the pleura and peritoneum are related to certain types of industrial work and perhaps also to neighborhood exposure. The association between lung cancer and asbestos was first found in the British textile industry. Merewether's inquiry showed an unduly high proportion of lung cancers at. autopsy in cases of asbestosis, and other observations among textile workers have confirmed the association.5-8 Two studies in particular which take some account of the degree of exposure suggest a dose-response relationship. Knox et al5 found a consider able excess of lung cancer in persons exposed before 1933 when asbestos control regula tions were enforced, but none in persons exposed for ten years or more in the most dusty areas since then. These findings must be reviewed when a longer period of obser vation has elapsed, since Newhouse8 (also in a textile factory) found an excess of lung cancer in persons heavily exposed 20 years earlier for periods of less than two years. Workers exposed to light or moderate dust concentrations showed no excess of broncho genic cancer even after 20 years of exposure. Selikoff and his colleagues9 found an eightfold increase in lung cancer compared with national figures in a cohort study among the members of an American insula tion workers union who had been exposed for 20 years or more. He also found a gross excess of pleural and peritoneal mesothe liomas. On the other hand, in a national survey of all known mesothelial tumors in Canada, 1960 to 1968, McDonald and her colleagues10 found a history of occupational contact with asbestos in a relatively small proportion of cases. This exposure was main ly in insulation and allied trades, rather than in the asbestos-producing industry. Of the four main types of asbestos fiber-- chrvsotile, crocidolite, amosite, and antho- phyllite, the two textile factories referred to above5-8 used mainly chrysotile fiber, togeth er with some crocidolite, and in one of the two, some amosite. As in most industrial applications different types of fibers are mixed, the carcinogenic effects of a single fiber type in practice can only be studied in mining and milling. Chrysotile fiber is of greatest importance because of its qualities and extensive usage. The only studies of lung cancer in chrysotile miners and millers were reported by Braun and Truan in 195S11 and by Kogan et al in 1966.13 In the first of these, nine deaths from lung cancer were observed in the Quebec industry com pared with six expected from provincial rates. The second was in the Soviet Union and showed that' compared with the general population the mortality from lung cancer was increased by a factor of 1.9 for miners, 3.1 for millers, and 2.3 for factory workers. The present investigation forms part of a comprehensive epidemiological survey of the entire Quebec asbestos-producing industry since its inception. Using the considerablevolume of data available, our primary ai has been to define as accurately as possible the quantitative relationship between expo sure to chrysotile asbestos and the incidence of lung cancer. The results from parallel studies concerned with the relationship of dust exposure to radiographic appearances, pulmonary' function, and respiratory symp toms will be reported separately. Materials and Methods Registration.--A register was compiled in the personnel department of each asbestos min ing company in the Eastern Townships regie:: of Quebec listing all persons currently or pre viously employed, as of Nov 1, 1966. A card was made out for each employee on which v.-.-.irecorded his name, date of birth, address, and a detailed work history which included the de partment and mine, dates of starting and finishing, for every job, and all periods of leave. In one of the 44 mining companies represent ed in our survey, the records of at least 560 persons had been destroyed. These were of ex-employees and others not transferred when ownership of the company changed in 1964. The work histories of a small number of older men who had been employed partly in this company and partly in others were also incom plete. Arch Environ Health--Vol 22. June 1971 I .I ASARCO ALV 0005733 MORTALITY AND CHRYSOTILE MIXES--McDONALD ET AL Table 1.--Subjects of Study and Results of Tracing by Age and Sex Year of Birth 1691*1895 1896-1900 1901*1905 1906*1910 1911-1915 1916*1920 All M 1.553 1.962 2.027 1.897 1.837 2.027 11,323 Persons F 14 31 50 99 117 154 465 Total 1.567 2.013 2.077 1.996 1,954 2.181 11.783 M 1.179 1.588 1.724 1.741 1.773 1.976 9.981 Traced F 13 28 45 92 113 149 440 Total 1.192 1.616 1.769 1.833 1.886 2.125 10,421 % Traced 76.1 80.3 85.2 91.8 96.5 97.4 88.4 Dead 611 655 498 330 221 142 2.457 %of Persons Traced 51.3 40.5 .28.2 18.0 11.7 6.7 23.6 Table 2.--Subjects of Study and Results ol Tracing by Dust Index and Years of Employment DustInoex Years <1 110* 30All No. % traced No. % traced No. % traced No. % traced No. % traced CO 3.043 76.6 1.025 86.3 55 100.0 0 4.123 79.9 10* 654 75.1 1.975 88.6 1.117 99.0 204 100.0 3.950 89.9 100- 38 76.3 450 e9.3 566 98.9 181 100.0 1.235 94.9 200* 3 66.7 264 90.5 522 99.0 250 1C0.0 1.039 97.0 4000 122 91.0 432 99.3 303 100.0 B57 98.4 8000 64 90.6 255 99.6 265 99.6 584 98.6 All 3.738 76.3 3,900 88.3 2.947 99.1 1.2C3 99.9 11.788 88.4 679 There was of course some movement of labor from one mining company to another. This necessitated bringing together and matching (he records of all the companies, to obtain complete work histories. In the course of trac ing ex-employees, the matching process could often be confirmed or amended. In all, 27,669 men and women were registered, including 6.415 currently employed on Nov 1. 1966. Of this total, 1,039 persons had been employed by more than one company. Using the occupational histories, a list was compiled of all named jobs in each company from the beginning of its operation. A descrip tion of each of the 13,346 jobs listed in this way was obtained from existing evaluation systems and by interviewing employees with long ser vice. Jobs with several names but involving the same work and dust exposure were combined and the codes finally needed for classification thus reduced to 5,783. For each of these and for each year of operation, the average dust expo sure was estimated on a 13-point scale. Throughout the industry in the dryers, crush ers, and mills, the concentrations of dust at working places had been measured annually since 1949 by Maurice Lachance. Eng. and since 1946 by various investigators in the small factory. All measurements were made with ei ther a midget impinger or Greenbure-Smith, impingor and rccordtd in millions of dust parti cles per cubic foot (mpcf). Levels of exposure in mining and maintenance jobs and in other operations where no dust measurements had been made were estimated by investigating present-day dust levels and adjusting these ac cording to changes in operating procedure re ported by mining and maintenance personnel. A full description .of geological and environ mental features and of the methods and results of dust measurement are to be published by G. W. Gibbs, MSc, and Mr. Lachance. Before 1949. few dust measurements were recorded, and estimates of exposure for that period are, therefore, only approximate. Inter views with employees of long service enabled comparisons of dustiness to be made with con ditions since 1949. The dates of installation or modification of dust control systems were also taken into account Duration of exposure for each person was expressed in years or fractions of years after subtraction of all periods of leave, and was corrected to a working week of 40 hours. A dust index was calculated for each employee by adding together the products of time spent on each job and estimated average dust concentra tion. For example, suppose the dust index for a man was 600: this might mean that he had worked for five years at 40 mpcf, 15 years at 20 mpcf, and 20 years at 5 mpcf tie, 200 + 300 + 100). It would be useful if it could be assumed Arch Environ Health--Vol 22. June 1971 ASARCO ALV 0005734 680 MORTALITY AND CHRYSOTILE MINES--McDONALD ET AL that an index of 600 were also equivalent to a working life of 40 years at 15 mpcf, but this would entail assumptions which may or may not be justified. Selection of Cohort.--The study of mortality was based on employees in an age group thought likely to yield the most valuable infor mation. The cohort selected comprised 11,788 persons born between 1891 and 1920 inclusive who had been employed for one calendar month or more. Those born'after 1920 would have had periods of exposure well short of a working lifetime and would still not have reached an age of high mortality. Those bom before 1891 would have been very difficult to trace, and, as mentioned above, some of their work histories were incomplete or missing. The selected cohort comprised among others 1,203 persons who had worked for 30 years or more, 3,738 who had worked for less than a year, and 1,080 who had worked for more than a year but whose job had not involved any appreciable exposure to dust. Thus, there was a substantial group with minimal exposure available for com parison with others who had been heavily ex posed. Tracing and Ascertainment.--The search for ex-employees was mainly initiated in the min ing towns. Telephone or postal inquiries were made first to establish whether the person in question was alive or dead on Nov 1, 1966. If reported dead, the exact date and place was sought from relatives, friends, and parish regis ters. For those not traced by these means, a systematic search was made in the provincial death records and certificates, and about 200 persons were found in this way. In addition, a search on our behalf was made in the index of the Canadian Unemployment Insurance Com mission for any person not found by means of local enquiries. The names of 598 persons were found who had registered or re-registered be tween 1964 and Nov 1, 1966, and 30 more had re-registered after that date. All 628 were pre sumed to be alive for the purpose of our study. Information was obtained concerning 10,421 (88.4%) of the 11,788 persons in the cohort: 7,965 were still alive (including 2.484 currently employed) and 2,457 (23.6% of those traced) were reported dead. Copies of death certificates were obtained from- Canadian Provincial and US State Registrars for 21211 (90%) persons. For 59 more (2%), an acceptable cause of death was obtained by other means: 30 were accidental deaths in the mines or during active service with the armed forces and 29 were caused by disease. Twenty-two of the 29 had m-curred before the Introduction of death regis tration in Quebec in 1926, and the cause of death was stated by relatives. In the other seven, the cause of death was obtained from a reliable source. A death certificate could not be obtained for the remaining 187; 47 had oc curred before 1926, and 100 were deaths outside Quebec, mostly outside Canada. The cause of death was coded according to the Seventh Revi sion of the International Classification of Dis ease (ICD [World Health Organization, Gene va, 1957]). This was done by senior coder of the Department of Demography of Quebec who had recently retired. The proportion of subjects traced by age, sex. and category of exposure is shown in Tables 1 and 2. Success in tracing depended mainly on duration of employment: 94% of those em ployed for one year or more and over 99% of all those employed for ten years or more were traced. Information was least satisfactory in persons bom before 1900 and in those em ployed for less than one year. The principal reasons for failure were (I) no relative or friend could be located in the neighborhood and (2) insufficient identifying information, which would have enabled us to make a request to the Unemployment Insurance Commission or to locate a death certificate in Quebec. There w, no other obvious reasons for bias towards a. covering the living or the dead, but those traced dead or alive are likely to have been long-term rather than transient residents. Diagnosis of Lung Cancer.--To make our figures for lung cancer as accurate as possible, we investigated and reviewed all certified cases and searched also for cases not described as such on the death certificate. Of 95 deaths coded as due to lung cancer (ICD 162 and 163), one was certified as due to a hydrone phroma and another as due to pulmonary fibrosis, both of which had been coded incor rectly. Inquiries about the remaining 93 cases were made from hospitals, physicians, and pr thologisls, and information was received in 70. In five of these it was reported that there had been an autopsy, but we were unable to obtain pathological details. In another five, though the diagnosis had been made in the hospital, no clinical or pathological record was found. Sev enteen more had been certified by family phy sicians but we did not succeed in finding out where and how the diagnosis had been made. In the remaining 48 cases, pathological re ports were obtained, 25 at autopsy and 23 at biopsy. A diagnosis of carcinoma of the trachea was made in one, of bronchus or lung in 44, and of malignant mesothelioma of pleura in two. In one case the histology was that of a melanosarcoma. and the pulmonary tumor was thought...,.o be metastatic. Thus, of 95 certified cases, Arch Environ Health--Vol 22. June 1971 ASARCO ALV 0005735 MORTALITY AND CHRYS0T1LE MINES--McDONALD ET AL 681 had been included incorrectly, 47 were con firmed pathologically, 28 hospital diagnoses were presumed correct, and in 17 no further information was obtained. The following steps were taken to discover cases of lung cancer that had not been men tioned as such on the death certificate: 1. A letter was sent to the certifying physi cians whenever it was stated that an autopsy had been or would be performed. These num bered 306 in all: 33 deaths were certified as due to lung cancer and 273, as due to other causes. In 18 replies, it was stated that there had been no autopsy: in 175, that the certified diagnosis was confirmed at autopsy; and in 18, that it was incorrect. These 18 yielded five additional cases of lung cancer. The diagnoses for these had been certified as renal tumor (ICD 180), infectious hepatitis (ICD 092), pulmonary fibrosis (ICD 519), diabetes (ICD 260), and asbestosis (ICD 523). 2. We asked the physicians in charge of the industrial clinics at Thetford Mines and Asbes tos for lists of all deaths known to be due to lung cancer. Their records included only em ployees who had given up work because of ill health, those who were pensioned, and those who had applied for workmen's compensation. They listed 24 of the '96 lung cancer deaths found by the methods already described plus three additional cases. These three deaths had been certified as due to tuberculosis (ICD 019). asbestosis (ICD 523). and coronary heart dis ease (ICD 420), and though there had been autopsies for the purpose of compensation, no mention of this was made on the death certif icate. In addition, there were three autopsies for compensation purposes in which a small lung cancer was noted as an incidental finding. The pathologist had not considered these can cers to have contributed to death and the causes certified were asbestosis (ICD 523) in two cases' and cardiac infarction (ICD 420) in a third. 3. A search was made for any case of prima ry malignant mesothelial tumor of the pleura that might have been given a code other than 162. One additional case coded as a benign pleural tumor was found. 4. We reviewed cases reported by patholo gists in the national survey of primary malig nant mesothelial tumors in Canada between 1960 and 1968, and five were in men who had worked in the Quebec asbestos-producing in dustry. Two were bom in 1922 and were, there fore, not in our cohort, and two died in 1967 after the present study had ended. A fifth case, diagnosed by biopsy, was eligible for both studies. To summarize the results of all these inquir ies, three cases were removed from the group of 95 malignant neoplasms of the bronchus, tra chea, or lung (ICD 162 and 163) and nine were added, making 101 in all (100 men and one woman). Included in the total were the follow ing: cancer of the trachea, in one case; malig nant mesothelial tumor, in three; and cancer of the bronchus or lung, in 97. Results Female Mortality.--There were only 465 women in the cohort; almost all had short exposures and all but 33 were in the two lowest dust-index categories. In all, 440 (94.6%) were traced, and of those, 45 had died. Tuberculosis (11), malignant neo plasms (12), circulatory (nine) and re spiratory (one) diseases, and trauma (one) were the causes of death in 34. Death was due to other causes in eight cases, and the cause of death was unknown in two: there was one death from lung cancer. Apart, perhaps, from tuberculosis, these figures are not unusual and will not be considered fur ther. Male Mortality.--To permit comparisons of mortality by dust index and years of expo sure, death rates were calculated in five-year cohorts by date of birth. Equivalent average death rates13 were then calculated using a standard population with equal numbers in each age group. Since the actual number at risk in each cohort was similar, this method of age standardization was particularly ap propriate. The standardized rates are shown by dust index and main disease groupings in Table 3. In the lower five dust categories, equivalent average death rates for "all caus es" were approximately the same, but in the top category, containing 5% of the total cohort, mortality was about 20% higher than the rest. Malignant neoplasms, circula tory diseases, and respiratory diseases, in approximately equal proportions, accounted for the excess. Two groups showed the oppo site trend; "trauma," probably because rates fell with advancing age, and ``unknown causes," probably because death certificates were difficult to find for older men with short periods of employment. Table 4 shows that the main contribution to the excess in malignant disease was from cancer of the bronchus, trachea, and lung Arch Environ Health--Vol 22. Jane 1971 ASARCO ALV 0005736 682 MORTALITY AND CHRYSOTILE MINES--McDONALD ET AL Table 3.--Equivalent Average Death Sates per 1,000 Men by Dust Index Oust Index No. of men Tuoerculosis (ICO 001*019) Malignant neoplasms (ICO 140*209) Circulatory diseases (ICO 400*469) Respiratory diseases (ICO 470*529) Trauma (ICO 800*999) Other causes Unknown causes All causes <10 3.006 18.8 38.8 85.0 9.5 34.7 46.S 30.6 264.1 10* 3.408 19.9 39.3 80.8 15.3 30.8 52.9 21.4 260.4 100* 1.148 31.6 32.2 82.5 19.5 36.3 39.0 16.5 257.6 2001.002 23.8 27.5 83.5 16.7 35.2 47.7 6.2 240.6 400* 842 28.5 45.1 94.8 15.3 33.3 38.3 7.2 262.9 800* 575 25.8 61.8 107.3 41.6 13.7 58.5 3.4 312.5 Alt 9.981 22.5 38.6 85.8 15.6 31.5 4S.4 20.8 263.1 Table 4.--Equivalent Average Death Rates per 1,000 Men lor Malignant Neoplasms (JCD 140*209) by Dust Index Oust Index Location Esophagus & stomach (ICO 150-151) Intestine & rectum (ICD 152-154) Other abdominal areas (ICO 155-159) 8ronchus, trachea & lung (ICO 162-163) Other malignant neoplasms Ail malignant neoplasms (ICO 140-209) <10 10.2 (32) 2.5 (8) 4.0 (13) 7.6 (25) 14.4 (47) 38.8 (125) 10* 4.5 (13) 4.9 (15) 3.0 (9) 6.6 (26) 18.6 (54) 39.3 (117) 100- 2.7 (2) 3.2 (3) 1.7 (1) 11.2 (10) 14.5 (id 32.2 (27) 200- 7.0 (5) 3.9 (3) 1.3 (1) 8.9 (8) 6.4 (5) 27.S (22) 400- 11.3 (8) 5.6 (4) 2.8 (2) 15.8 an 9.5 (6) 45.1 (31) BOO* 13.7 (8) 8-7 . () 1.7 (1) 24.2 (14) 13.5 (8) 61.8 (36) All 7.5 (68) 4.0 (38) 2.5 (27) 9.9 (94) 14.4 (131) 3S.6 (3 5S> i'ICD 162 and 163). In this group there was little difference between rates in the four lower dust categories, but the fifth and sixth groups showed an upward trend. A similar trend with years of exposure was found (Table 5), but the figures within the body of the table suggest a closer relationship with dust than with years. Subtraction of the three incorrectly coded cases and addition of the nine lung cancer cases found at autopsy increased slightly the rate among persons with highest dust and longest exposure, but did not materially change the picture (Ta ble 5). Further analyses to distinguish bet ter the relative importance of years of expo sure and dust index are described as follows. Rates for cancer of the intestine were about one third of those for cancers of the bronchus, trachea, and lung, but showed a very similar trend. Rates for cancer of the esophagus and stomach, on the other hand. did not appear related to dust in any consis tent manner. Other abdominal neoplasms were less frequent still and also unrelated to dust. This is important since it might be expected that unrecognized peritoneal meso theliomas would be found within this grou- Most of the excess mortality from re spiratory disease was ascribed to pneumo coniosis (Table 6). There was little evidence that dust-associated deaths were included other respiratory categories. Of the 2S deaths coded under the pneumoconioses, one was described as anthracosis, four as silico sis, and the remaining 23 as asbestosis. The greatest excess mortality from pneumocon iosis was among persons in the highest dust group who had been employed between ten and 29 years. In the circulatory diseases group, there was also excess mortality in the highest two dust categories, mainly in persons emplo^-yi1.. Arch Environ Health--Vol 22. June 1571 t ASARCO ALV 0005737 MORTALITY AND CHRYSOTILE MIXES--McDOSALD ET AL 683 between ten and 29 years. It was present equally in the '`arteriosclerotic and degener ative heart disease" group (ICD 420 to 422 [which included more than three quarters of all circulatory deaths]) and in the group of "other circulatory diseases" (.ICD 400 to 419 and 423 to 469). Comparison With Quebec Mortality.-- The number of deaths from all causes which would have been expected if Quebec death rates had applied was calculated in the fol lowing way. Age-specific death rates for the province were applied for each year, 1950 to 1966, to all the men traced who were alive in 1950, with an adjustment for those who started work after 1950. The expected num ber of deaths among men, thus, was 1,824, whereas the observed number in the cohort was 1,674. A similar calculation was made for lung cancer deaths. The expected number was 91, whereas 94 male deaths certified as due to this caus4 were observed. Correction for cod ing errors and additional autopsy informa tion was not appropriate, since certificates for the general population were uncorrected. In the six counties of the province which include the mining region lArthabaska, Beauce, Drummond, Megantic, Richmond, and Wolfe), the lung cancer death rate was about two-thirds the provincial rate and the expected number of deaths was correspond ingly lower. However, many ex-employees were no longer living in this area when they died, and most of those who had moved had gone to cities where lung cancer death rates were higher. The best estimate of expected deaths probably lies between the numbers 61, derived from the mining region, and 91, from the province. The excess of observed over expected lung cancer deaths, therefore, lies somewhere between zero and 30 and is most probably between ten and 20. Factor Evaluation.--Although the num bers of men in each cohort were about the same, their distribution by years of employ ment and dust index was uneven. In these circumstances, equivalent average death rates could be misleading. Berry1-' recently reviewed some of the methods used for ana lyzing the importance of factors in multiway tables and described a parametric approach which has the advantage that the adequacy of the model may be checked and statistical Table 5.--Equivalent Average Death Rates per 1,000 Men tor Cancer ot Bronchus, Trachea, and Lung* Oust Index Year* <10 10- 100- 200- 400* 600* All As certified <i 6.4 (16) i- 11.7 (9) 10* 0.0 (0) 30- ... All 7.6 (25) 1.7 (1) 9.5 (15) 12.4 (7) 12.5 (3) 8.6 (26) 27.8 (1) 11.3 W 6.7 (3) 14.7 (2) 11.2 (10) 0.0 (0) 5.8 (1) 5.3 (3) 13.9 (4) 6.9 (8) Using ail availacie nrtfcrmation <1 6.7 1.7 0.0 0.0 (17) (i) (0) (0) 1- 12.0 10.1 9.4 11.5 (9) (16) (3) (2) 10- 0.0 12.8 7.6 9.2 (0) (8) (4) (4) 30- 12.5 14.7 13.9 (3) (2) (4) All 7.8 9.1 10.1 10.6 (26) (28) (9) (10) _ 7.6 (1) 19.9 (6) 16.1 (4) 15.8 (11! 7.6 (1) 19.9, (6) 16.1 (4) 15.3 01) ... 0.0 (0) 23.6 (6) 27.9 (8) 24.2 (14) 5.9 (18) 9.6 (30) 13.1 (25) 17.2 (21) 9.9 (94) 5.9 , (IS) 0.0 10.1 (0) (31) 23.6 14.3 (6) (22) 35.3 15.1 (10) 123) 27.7 1C.6 0 6) (ICO) 9y dust index ana years of employment. Diagnosis was ICD 162 and 163. significance evaluated. From the number of deaths from any particular cause in groups subdivided by date of birth and year of employment or dust index, expected rates may be calculated assuming no interaction between age and the other two factors. For this analysis, the complementary log log transformation was used.13 This transforma tion is appropriate when increasing expo sure is associated with proportional in creases in the age-specific death rates. The results of this parametric analysis for respiratory cancer '.Table 71 agree remark ably closely with the equivalent average death rates shown in the lower half of Table 5. The rates for the first five dust index levels do not differ significantly but those for the first four are significantly different from the highest exposure group. There are, however, also differences between the rates by years of employment, those for the lower two categories being significantly less than that for men with the longest exposure. The general fit of the complementary log log model is very good. In no cell of the 24 does the observed number of deaths differ significantly from expectation. Tests for in teraction between the 24 cells and the six cohorts gave the following results: likelihood Arch Environ Health--Vo! 22. June 1971 II ASARCO ALV 0005738 684 MORTALITY AMD CHRYSOTILE MIXES--McDOXALD ET AL V.: Table 6.--Equivalent Average Death Rates per 1,000 Men tor Respiratory Diseases (ICD 470-529) by Dust Index and Years ol Employment Dust Index Years <10 10- 100- 200- 400* Pneumonia and bronchitis (ICO 490*509) <1 2.6(7) 8.7(4) 0 (0) i- 7.1(5) 3.7(6) 2.5(1) ...10- 0 30- 6.3(4) 0 5.9(2) 4.9(1) 166.7(1) 4.4(1) 5.3(3) 5.8(1) ... 15.5(2) 5.6(3) 0 (0) All 3.8(12) 4.7(14) 4.2(4) 6.6(6) 5.6(5) Pneumoconioses (ICD 523 and 524) <1 1.4(3) 4.5(2) 0 1* 1.4(1) 1.5(2) 0 10- 0 2.3(1) 0 30- 0 0 0 5.8(1) 1.2(1) 3.6(1) ... 0 1.6(1) 7.1(2) All 1.5(4) 2.0(5) 0(0) 3.7(3) 2.9(3) Other respiratory diseases (ICD 470*489. 510*522, 525*529) <1 4.5(10) 5.7(3) 0 00 1- 3.8(3) 13.2(20) 29.0(9) 9.3(2) 7.9(1) 10* 0 30* 2.7(2) 0 9.2(2) 0 7.9(2) 0 11.0(3) 0 All A.?n3) *'.6(25.) 15-3(11) 6.3(4) 5.9(4) All respiratory diseases (ICD 470*529) <i 8.6(20) 18.8(9) 0 1- 12.2(9) 18.4(28) 31.5(10) 10* 0 11.2(7) 15.1(4) 30* 0 4.9(1) . ( )Ait 9.5(29) 1 0 3 44 19.5(15) 166.7(1) 20.0(4) 14.4(6) 9.3(2) 16.7(13) 23.5(3) 18.2(7) 7.1(2) 15.2(12) 800* 16.7(1) 9.5(2) 3.4(1) 6.9(4) ... 0 34.3(8) 17.9(5) 22.6(13) 0 13.9(1) 7.8(2) 13.5(4) 12.1(7) ... 30.6(2) 51.6(12) 34.7(10) 41.6(24) All 4.0(12) 5.2(16) 6.3(14) 2.9(3) 4.9(45) 1.9(5) 1.5(4) 6.3(11) 6.9(8) 3.2(28) 4.6(13) 12.5(26) 7.2(11) 3.6(4) 7.5(64) 10.6(20) 19.2(56) 19.9(36) 13.4(15) 15.6(137) ratio test, 88.9 and x:> 90.4, each with 95 degrees of freedom; and P > 0.5. Comment Table 7,--Age-Corrected Death Rates per 1,C' Men tor Cancer of the Bronchus, Trachea, ai. Lung by Dust Index and Years ol Employment *" Oust Index At face value, the findings suggest that our cohort of workers in the chrysotile min ing industry had a lower mortality than the population of Quebec of the same age. This is generally true of employed persons, pro vided they are not subjected to an occupa tional hazard sufficient to offset the con siderable selective advantage of being and remaining fit for work. This advantage was clearly lost by the men in the highest dustindex category whose standardized mortali ty was about 20% above that of the rest. Two thirds of the excess mortality in this group was probably due to pulmonary fibro sis, shown on the death certificate as either asbestosis or in the guise of various respira tory or cardiovascular diagnoses, and the remaining third to cancer, mainly of the respiratory tract. The high rate of lung cancer in men heavily exposed to asbestos might be ex plained if such men also tended to smoke more heavily than others. We have no direct evidence on this point for ex-employees, since information on smoking was not rou- Years <1 1* 10* 30* All <10 6.3 11.4 0.0 ... 7.5 10* 2.0 9.4 10.4 13.3 8.7 100* 0.0 8.1 9.8 9.9 8.9 200* 0.0 8.3 10.3 14.6 11.0 4.0.0. * 9.6 16.8 13.7 14.6 800* 0.0 23.4 34.4 26.3 All 5.5 9.5 12.8 17.9 10.0 * A complementary log Jog mocel was used, anc ciag* nosis was based on all available information. tinelv recorded by the companies or the medical clinics. However, in a survey of a selected random sample of over 1,000 cur rent employees there was little or no rela tionship, after allowing for age, between smoking habits and either dust exposure c: duration of employment. Our attempts to assess separately the importance of cumulative dust exposure and duration of exposure in relation to lung cancer are capable of more than one inter pretation. As shown in Table 5, there is a fivefold difference between the mortality for those with the lowest amount and duration of exposure, 6.7, and that for those with the highest, 35.3. This is confirmed by the para metric analysis (Table ?' which further sug gests that accumulated dust exposure...:v;.rd Arch Environ Health--Vol 22, June 1971 ASARCO ALV 0005739 MORTALITY A\'D CHRYSOTILE MIS'ES--McDOS'ALD ET AL 685 Table 8.--Comparison of Mesothelial Tumor Findings in Two Previous Studies and the Present Study No. of men studied No. of deaths All causes lung cancer Mesothelial tumors Selikoff et al'< 632 380 72 22 Newhouse*-17 4.806 436 42 20 Present Study 9.981 2.413 97 3 duration of employment are about equally important in determining the difference. Un fortunately, there are possible sources of bias and error which may have contributed to both these effects. In Table 5, the rates for men with less than one year of employment seem remark ably low; 18 deaths were observed, whereas 29 would have been expected on the basis of Quebec rates. It was in this group that tracing was least satisfactory (Table 2). Failure to ascertain even a small number of lung cancer deaths would have made an appreciable difference in the rates. Short term employees who were traced tended to be those who had stayed in the neighbor hoods where, in any case, lung cancer rates were much lower than in the urban areas of the province and elsewhere, to which those untraced may well have gone. Whatever the explanation, it is difficult to accept, without reserve, the low rate of lung cancer in this group. A second source of error is related to the ascertainment of lung cancer as a cause of death. It is generally believed that this diag nosis is greatly aided by postmortem exami nation. In our study, 34% of the cases be fore correction and 40% after correction had had an autopsy, compared with 12% of all deaths. This might not matter if autopsies were evenly distributed in relation to expo sure, but this was not so. The autopsy rates in ascending order of dust-index group were 11%, 11%, 9%, 10%, 16%, and 22%, and by years of employment, 12%, 9%, 17%, and 21%. This trend is also likely to have exaggerated the difference in lung cancer rates in relation to exposure, but by how much it is impossible to say. Taking all these considerations into ac count we are inclined to conclude that the true difference between those maximally and minimally exposed may well be closer to threefold than fivefold and that this is part ly dust-related and partly time-related. We propose in future analyses, when a longer period of observation will have yielded more data, to use an exposure index based on the concept of amount of dust inhaled and the time that it remains in the lung. Our findings so far appear compatible with such a model. In the Canadian survey mentioned ear lier,10 primary malignant mesothelial tumors were rarely associated with chrvsotile asbes tos production in Quebec and the present survey bears this out. Three deaths from this cause were found among nearly 2,500 deaths from all causes in the cohort. This is probably more than would be expected in a comparable number of deaths in the general population, but quite out of line with the findings of Selikoff et al10 in insu'ation workers and these of Nev.'house'17 in a Lon don asbestos factory. The magnitude of the difference can be inferred from the figures in Table 8. Though these figures are not entire ly comparable, because of age, methods of ascertainment, or other factors, they suggest that the results of Newhouse517 are similar to those of Selikoff et ai18 for mesothelial tumors and, perhaps, lung cancer. It is clear that the Quebec chrvsotile workers have had nothing like the experiences of the America-1 insulation workers or the London factory workers with respect to malignant mesothe lioma, and it seems unlikely that they are compatible with respect to lung cancer. These findings strongly suggest either that chrvsotile is less likely to cause malignant disease of the lung and pleura than other forms of asbestos, such as crocidolite, or that workers engaged in insulation and process ing are exposed to additional factors which explain the difference. This work was undertaken with the assistance of a grant from the Institute of Occupational and Environmental Health of the Quebec Asbestos Min ing Association. Paul M. O.utier, MD. medii'nl director of the Arch nu<ron Health--Vof 22. June 1971 i .1 ASARCO ALV 0005740