Document dYnk47vrMaQGr5qyGpgev4jqG

MORTALITY EXPERIENCE OF INSULATION WORKERS IN THE UNITED STATES AND CANADA. 1943-1976* Irving J. Seiikoff Environmental Sciences Laboratory Mount Sinai School of Medicine The City University ofSew York Sew York, Sew York 10029 E. Cuyler Hammond and Herbert Seidman Department ofEpidemiology and Statistics American Cancer Society Sew York, Sew York 10017 PLAINTIFF'S EbAXIHMIVBIIT > 54-105* In 1964. we reported data concerning the mortality experience of a cohort of asbestos insulation workers in the New York-New Jersey metropolitan area 1943 1962.* Significant increases in deaths of lung cancer, mesothelioma, gastrointestinal cancer and asbestosis were found. Findings were limited to observation of only 632 men. however. On January 1. 1967. a larger cohort was enrolled for prospective observation and included all 17,800 men on the rolls of the insulation workers' union (International Association of Heat and Frost Insulators and Asbestos Workers. AFL-CIO, CLC) in the United States and Canada on that day. This second cohort included the survivors of the original 632 New York-New Jersey men. Observation has been maintained and an interval report of deaths in this group since January I. 1967 has been made.2 The present analysis details findings among these men through December 3J, 1976. Asbestos Insulation in the United States Asbestos, as a mixture of fiber ana sodium silicate, was first used as an insulation material in 1866 and as asbestos cement about 1870. Magnesia with asbestos as a binder soon followed and air-cell covering, using corregated asbestos paper, was introduced in 1898. The first union of insulation workers in the United States was formed in New York City in 1883 under a charter issued by the Knights of Labor (predecessor of the American Federation of Labor), as "The Salamander Association of Boiler and Pipe Felters." The present union of insulation workers in this country as chartered by the AFL in 1910, as the International Association of Heat and Frost Insulators and Asbestos Workers, by the amalgamation of the Salamanders Association of New York with other independent locals throughout the country. Members of this union are insulation workers, primarily employed in the building trades doing construction insulation work but also employed as-insulation workers in refineries, industrial plants, This study was supported by grants from the American Cancer Society (R-53), the National Institute for Occupational Safety and Health (0HOO32O) and the National Institute of Environ mental Health Sciences (ES00928) of the U S. Department of Health, Education and Welfare, and the Health Research Council of the City of New York (U1272). 91 0077-8923/79;0320-0091 SI.75/0 C 1979, NYAS v Annals New York Academy of Sciences shipyards and powerhouse construction and repair. Much of their work is in the open but sometimes, as in shipyards, in rather tight quarters. The men generally work in ail parts of the trade, with few specialists at one or another part of it. Conditions at work vary from job to job and from company to company, with fewer than 10le of the men remaining with one company during their working lifetimes.3 Data published concern ing work practices in other countries suggest that asbestos exposure in insulation w'ork has been approximately the same the world over.4 Asbestos exposure to which insulation workers are subject has been limited, varied and intermittent. Some of the materials used have contained no asbestos. In the early part of the century, magnesia pipe coverings used shredded rope as a binder, with no asbestos. Somewhat later, wooi-feit. cork and rock wool were used. Starting during World War II. fibrous glass products came into increasing use and. during the 1970s. their utilization was greatly expanded, as asbestos content of insulation materials sharply decreased. Magnesia biock insulation, including molded pipe coverings, usually contained approximately 15^ asbestos, while asbestos cement generally had 15-2011 or less of asbestos. Asbestos paper products contained a higher asbestos content, but were used in much smaller quantities. Until approximately the early 1940s, chrysotile alone was utilized in the manufac ture of the asbestos insulation products used by these men. Amosite began to be used in the mid- 1930s in small quantities but became more widely utilized during World War II and subsequently/ Few dust counts were made in insulation work until the mid-1960s. However, analysis of available data, including reconstruction of work situations and extrapola tion to the past of observations made more recently, suggest that insulation workers would have been exposed to dust levels of 4-12 fibers/ml (as time weighted Table 1A Membership of New York and New Jersey Insulation Workers* Union* January 1, 1943. Classified by age and Years from First Exposure to asbestos Dust Total Number Age. of Number of Years Since First Exposure to Asbestos Years Members 0-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50 r- 15-19 22 20-24 14 14 25-29 38 19 19 30-34 122 2 94 26 35-39 40-44 76 74 I 2! 47 6 1 1 4 28 30 10 1 45-49 77 1 19 29 24 4 SO-54 82 1 2 9 24 18 24 4 55-59 65 1 4 9 li 20 19 1 60-64 41 2 5 8 3 13 6 4 65-69 70-74 23 13 1 2 2 1 3 8 42 2 2 2 2 23 75-79 3 3 80-84 2 11 Total 632 40 145 140 108 70 67 40 15 4 3 'Locals 12 and 32 of the International Association of Heat and Frost Insulators and Asbestos Workers, AFL-CJO. - C_\SS:.-:=: A::j; Agt ' <<_ 60-< 65-' SO85 To:. Interne:; averages).' also have c- On Jan of the Inter the New Y tion accord observation Each o; deaths are employmen who achie1* experienced onset of cm are not com more clear; that there observation excess was effect."1 Observby Deccrr.b. of causes of :< is in the open ally work in all .aitions at work ; 0% of the men fished concerninsulation work limited, varied os. In the early binder, with no Starting during .ring the 1970s, ation materials pipe coverings, t generally had nigher asbestos n the manufacigan to be used 1 during World >60s. However, and extrapol ation workers time weighted >s Dust stos 4 45-19 50 + 2 23 43 rs and Asbestos Seiikotf et al.\ Mortality of Insulation Workers 9} T.A8LE IB Man-years of Observation, January 1. 1943-Decsmber 31. 1962. OF January 1. 1943 Membership of New- York and New Jersey Insulation Workers' Union* Classified by Attained Age and Years from First Exposure to asbestos Dust Attained Age. Years 15-19 20-24 25-29 30-3-4 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85aTotal Total Man-years 4 44 165 542 1,068 1.339 1.588 1.562 1.344 1.142 849 513 242 84 29 10.515 <20 4 44 165 542 773 23 74 39 :s 22 17 9 -- -- -- 19`0 Years from Onset 20-34 -- -- -- -- 295 1086 1510 1368 782 361 175 72 33 1 -- 5683 -- -- -- -- -- -- 4 1:5 *34 759 657 l 209 83 29 2862 International Association of Heat and Frost Insulators and .Asbestos Workers. AFL-CIO. averages).5 While there might have been periods of little or no exposure, there could also have been times of peak exposures much higher than the calculated averages. Mortality Experience oe New York-New Jersey Insulation Workers 1943-1976 . On January 1, 1943, there were 632 men registered as members of the two locals of the International Association of Heat and Frost Insulators and Asbestos Workers in the New York-New Jersey metropolitan area (Locals 12 and 32). Their age distribu tion according to years from onset is contained in Table 1A and man-years of observation 1943-1962 are shown in Table 1B. Each of these men was traced to December 31, 1962: expected and observed deaths are found in Table 2. Nine men died before reaching 20 years from first employment, and they are excluded from Table 2. which is limited to the 623 men who achieved that point. While deaths related to asbestos exposure of the kind experienced by insulation workers may sometimes occur in less than 20 years from onset of employment (lung cancer, asbestosis and. occasionally, mesothelioma) these are not common and we consider that analyses of experiences beyond the 20-year point more clearly define the influence of such exposure. It was of interest, parenthetically, that there was no excess in total deaths of all causes during the first 10 years of observation, 1943-1952. in the post 20 years from onset experience-(although an excess was seen for lung cancer specifically), as example of the "healthy worker effect.*'1 Observation of the survivors of the original cohort has continued prospectively and by December 31,1976.478 of the original 632 men had died. The same overall pattern of causes of death has continued' although distribution of deaths by cause has changed 94 Annals New York Academv of Sciences Table 2 Expected and Observed Deaths among 623 Asbestos Insulation Workers New York-New Jersey. 20 or More Years After Onset of Work January I. 1943-December 31. 1962 (8545 Man-years of Observation) Underlying Cause of Death Total deaths-ail causes Total cancer-all sites Cancer of lung Pleural mesothelioma Peritonea! mesothelioma Cancer of esophagus, stomach, colon-rectum Cancer of larynx, pharynx, buccai cavity Cancer of kidney All other cancer Noninfectious pulmonarv diseases, total Asbestosis All other causes Expected* 195.4 32.1 6.0 t t 9.7 1.7 0.7 U.O 4.0 t 159.3 Observed 253 95 42 3 4 29 2 0 15 14 12 144 Expected deaths are based upon white male age-specific U-S. death rates of the U.S. National Center for Health Statistics. 1949-1962. Rates for specific causes of death for 1943-1948 were extrapolated from rates for 1949-5955. tRates are not available, but these have been rare causes of death in the general population. somewhat, reflecting a number of epidemiological influences. Thus, pleura! and peritoneal mesothelioma, which tend to occur somewhat later than bronchogenic carcinoma, became proportionately more common (Table 3). This change in prepon derance also reflects the smaller proportions of older men who ever smoked cigarettes, and also a "survivor effect." Since the cigarette smokers in the original group had Table 3 Expected and Observed Deaths among 632 New York-New Jersey Asbestos Insulation Workers January I. 1943-December 31. 1976 (13,925 Man-years of Observation) Underlying Cause of Death Total deaths, all causes Total cancer, all sites Cancer of lung Pleural mesothelioma Peritoneal mesothelioma Cancer of esophagus Cancer of stomach Cancer of colon-rectum Cancer of larynx, pharynx, buccal cavitv Cancer of kidnev All other cancer Noninfectious pulmonary diseases, total Asbestosis AH other causes " Expected* 328.9 57.0 13.3 t t 1.4 5.4 8.3 2.8 U 24.5 9.3 t 262.6 Observed 478 210 93 n 27 1 19 23 6 2 28 45 4! 223 Expected deaths are based upon white malt age-specific U.S. death rates of the U.S. National Center for Health Statistics, 1949-1976. Rates for specific cause of death for 1943-5948 were extrapolated from rates for 1949-1955. tRates are not available, but these have been rare causes of death in the general population. ` cea C Si .-N<r mes UQU asbc into. repa Am. lung tiouand all s neop 49Tr of m 31 st of m pleu unler evali tant! from or cr. I work Eig'r. loca! had : obse: (Tae strau work not c expo: obscr expcr . and genic eponettes. o had V.S. :h for increased mortality risk (especially from lung cancer a.nc cardiovascular disease) there would iikeiy have been comparatively fewer men with a history of cigarette smoking and still fewer who continued smoking at least the same amount, among the cohort survivors, as the years went by. Except as influenced by other factors associated with advancing lapsed time since onset of asbestos work, this would make for fewer deaths of lung cancer, with more men at risk of dying of other asbestos-associated disease. Lung cancer remained the most important cause of excess deaths, with 93 such deaths observed vs. 13.3 expected. Thirty-eight deaths occurred of mesothelioma. 11 pleural and 27 peritoneal. The increase in gastrointestinal cancer originally reported in 1964 was again seen, with 43 deaths observed. 15.1 expected. There were 41 deaths of asbestosis. In addition, 4 deaths were observed of other noninfectious pulmonary diseases. Apart from cancer, asbestosis. and other noninfectious pulmonary diseases, observed deaths were fewer than expected of all other causes; only 223 seen. approximately 40 fewer than anticipated. . This experience, with 19% of deaths due to lung cancer. 8% the result mesothelioma, 9% of gastrointestinal cancer and 9% of asbestosis and other noninfec- lious pulmonary diseases, begins to provide a broad outline of the total mortality experience of insulation workers employed under conditions of the past (much less asbestos insulation was used in new construction after 1972. although variable and intermittent asbestos exposure would be expected to have continued to occur during repair work). 76% of the original cohort enrolled in 1943 had died by the end of 1976. Variations in distribution of deaths by cause over time may be seen in Table 4. Among the 9 deaths listed before 20 years from onset of exposure, there was none of lung cancer, mesothelioma, gastrointestinal cancer, or asbestosis and other noninfec tious pulmonary diseases. Variations are also evident in the two periods. 20-34 years and 35 or more years from onset. Although the proportion of deaths due to cancer of all sites was virtually identical in the two periods (44.5% and 44.3%). the specific neoplasms were quite different in proportional distribution. Lung cancer accounted for 49% of the cancer deaths in the shorter period but only 43% later. There were 7 deaths of mesothelioma in the 6263 man-years of observation during the 20-34 year span but 31 such deaths in the 5692 man-years later on. It is of interest, too, that the proportion of mesotheliomas that were pleural in location also changed over time; 4 of the 7 were pleural in the earlier period, against only 7 of 31 in the second period. It is clear that unless opportunity for prolonged observation is available, it will be difficult to fully evaluate the distribution of deaths by cause among asbestos-exposed groups. Concomi tantly, such analysis will be enhanced by considering deaths in each period of duration from onset, separately. Not unexpectedly, deaths of asbestosis were again largely concentrated in the 35 or more years from onset period; 38 of the 41 such deaths were found at this time. It has been of interest to inquire concerning the experience of those insulation workers whose exposure began after 1943. since this would reflect postwar conditions. Eight hundred and ninety men joined the New York-New Jersey metropolitan area locals of Lhe union during the period January I, 1943-Deeember 31, 1962. Most (833) had had no prior asbestos employment before entering the union. These men have been observed prospectively from the time of admission to the union to December 3!. 19"6 (Tables 5A and 5B). Our experience with their pre-1943 predecessors had demon strated that we were to expect relatively few deaths before 20 years from onset of their work. Still, because of the importance of evaluating postwar experience among men not exposed in earlier years, we undertook this study. Table 6 details their mortality experience from first employment to December 31, 1976. In the 15.520 man-years of observation during the less than 20-year period, there was no unusual mortality experience. Altogether, there were fewer deaths chan expected (again, the "healthy 5 D3 .1a-- t/"i __ - oI Se CO J^S O--' o *3 c c 'J 5 ii W *- j) 1*1 MS :|s 75 = S 55 -= * % ^ 90 ^ eJ fl* ; UV l* Table 5A New York and New Jersey Insulation Workers Observed January 1. I9-J3December 31. 1962. and Subsequently to December 31, 1976 (Number of Men Attaining Category) Period 1943-1952 1953-1962 1963-1969 1970-1976 Union Members Jan. 1. 1943 Years from Onset Total <20 20-34 35- 632 325 523 190 542 40 393 321 370 -- 154 328 234 -- 32 224 Men Joining Union 1943-1962 Years from Onset Total <20 20-34 35- 412 412 ___ ___ 833 833 -- -- 828 828 258 -- 811 561 515 -- Table 5B Man-years of Observation of New- York and New Jersey Insulation Workers. 1943-1976 (Number of Man-years) Period 1943-1952 1953-1962 1963-1969 1970-1976 Union Members Jan. 1. 1943 Years from Onset Total <20 20-34 35 + 5928 4587 2088 1322 1682 288 -- -- 3253 2430 418 162 993 1869 1670 1160 Men Joining Union 1943-1962 Years from Onset Total <20 20-34 35- 1658 5906 5752 5485 1658 5906 5156 2800 ___ -- 596 2685 ___ -- -- -- Table 6 Expected and Observed Deaths among 833 New York-New Jersey .asbestos Insulation Workers First Employed January 1. 1943-December 31. 1962. and Observed from First Employment-December 31. 1976 (Duration from Onset of Employment) Number of Men Attaining Category Man-years of Observation Underlying Cause of Death Total deaths, all causes Total cancer, all sites Cancer of lung Pleural mesothelioma Peritoneal mesothelioma Cancer of esophagus, stomach, colon-rectum Cancer of larynx, pharynx, buccal cavitv Cancer of kidney All other cancer Noninfectious pulmonarv diseases, total Asbestosis All other causes Less than 20 Years 83 3 15.5 20 Expected* Observed 39.8 5.1 1.1 t t' 23 5 2 0 0 20-34 Years 523 3281 Expected* Observed 24.S 39 5.0 15 1.8 8 t2 t1 0.7 1 0.8 2 0.2 1 0.3 1 0.1 0 0.1 1 3.0 1 2.0 0 0.5 0 0.6 7 t 0t 6 34.2 18 19.2 17 Expected deaths are based upon white male age-specific U S. death rates of the U.S. National Center for Health Statistics, 1949-1976. Rates for specific causes of death for 1943-1948 were extrapolated from rates for 1949-1955. tRatcs are not available, but these have been rare causes of death in the general population. rv*. v--:. r -+>- it> 98 Annals New York Academy of Sciences worker effect") and there was no increase in cancer deaths. No mesotheliomas were seen, nor deaths of asbestosis. There was no significant increase of either lung cancer or gastrointestinal cancer. Since the 20-year point, we have begun to see the changes predicted by our earlier studies. In the 32S! men-years of observation 20-34 years from onset, there were approximately three times as many cancer deaths as expected, primarily due to lung cancer. Mesotheliomas were now first seen and some deaths of asbestosis occurred. These had occurred primarily during the period 1970-1976 (Table 7). When the post-1943 workers were compared with those first exposed before that time, and evaluation was limited to observations less than 20 years from first exposure, it was found that neither group had unusual iung cancer, mesothelioma or asbestosis experience. Table 8 demonstrates that among 325 pre-World War 11 men who had not reached 20 years from onset of exposure during their period of observation, covering 1970 man-years, no excess of these diseases was seen. Similarly, among the 833 post-war men, with 15.520 man-years of observation during the shorter than 20-year from onset point, and with exposures starting 1943-1962, neither mesothe lioma nor asbestosis deaths occurred and there were not significant excess deaths of lung cancer. This is of interest in that, by and large, only chrysotile exposure had occurred before 1943 for the first group, while amosite was added to chrysotile in terms of potential exposure from 1943 on. The after-1943 group was potentially exposed to both types of asbestos throughout their work experience. These data indicate that amosite did not shorten the period of clinical latency for asbestosassociated disease in these insulation workers. Further, taken together with the data in Tables 6 and 7, these observations suggest that the mortality experience of insulators exposed to both chrysotile and amosite will be in the same direction as those originally exposed only to chrysotile and later to insulation dusts also containing amosite. Table 9, reviewing deaths of selected causes 20-34 years from onset in the post-1943 group, and comparing the findings with observations in the pre-1943 group for the same time span, suggests that this will be the case. Mortality Experience of Insulators in the United States and Canada 1967-1976 There were 17,800 men on the rolls of the asbestos workers union in the United States and Canada on January 1. 1967*. A good deal of information concerning these men was available, including date of birth, date of first insulation work, employment status on January l, 1967. Additional information was obtained by questionnaire from a majority of the men concerning current symptoms, respirator use, smoking habits, work practices. We have maintained observation of this cohort since 1967, with the valuable assistance of the local and international officers of the union. The men are registered in approximately 120 local unions in the various parts of the United States and Canada, including Alaska and Hawaii. Whenever an insulation worker associated with the union dies, we are notified. In most instances, a death certificate is forwarded: if not, it is obtained. As with the New York-New Jersey group, information is then sought concerning the details of the circumstances of death.1'' including clinical data, roentgenograms, histological material obtained at surgery or autopsy. The clinical and International Association of Heat and Frost Insulators and Asbestos Workers. AFL-CIO, CLC. T aih ii 102 Annals New York Academy of Sciences Table 10 Membership of asbestos Insolation Workers' Union*, January 1. 1967, Classified by Ace and by Years from First Exposure to Asbestos Dust Years 15-19 20-2-4 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75-79 80-84 85Total Total Number of Members 244 1.695 2.412 2.762 2.988 2.260 1.589 1,297 984 703 419 255 11! 52 29 17,800 0-9 244 1695 2066 1065 313 79 49 27 13 1 -- -- -- -- 5552 Nu mber of Years Since First Exposure to Asbestos 10-14 15-19 20-24 25-29 30-34 35-39 40--i9 345 1 1356 341 1141 1342 192 424 1026 591 139 l 131 433 442 487 47 88 214 332 377 182 77 49 129 206 176 146 193 72 21 59 131 126 87 99 179 6 18 41 58 45 29 201 -- 6 14 22 21 16 105 1 .-- 4 8 4 7 37 ------ 21 2 16 27 3562 3569 1953 1395 534 425 617 *0 -- 21 71 50 31 20 193 Membership in the United States and Canada of the International Association of Heat and Frost Insulators and Asbestos Workers. AFL-CIO. CLC. roentgenological data are reviewed and the pathological material examined. Most of the presentation for this cohort is in terms of underlying cause of death categorized according to best evidence available (BE). Further details are also given in the tables for underlying cause assigned on the basis of death certificate information only (DC). At the outset of the study, a majority of the men were below the age of 40 (10,101 of 17,800). The majority, too, had yet to achieve 20 years from first exposure (12,683 of 17.800) (Table 10). During the period January 1. 1967-December 31, 1976, the cohort changed its age and duration from onset distribution with the passage of time. Many of the men who started in the less than 20 year from onset group achieved 20+ years from onset during the decade of observation (Table 11). Altogether, 12,683 men, with 89,462 man-years of experience, suffered 325 deaths in the period before 20 years. There Table 11 Observation of 17,800 Asbestos Insulation Workers in the United States and Canada January 1. 1967-DECEM8ER 31, 1976 Number of men Man-years of observation Deaths Average age during observation Total 17.300 166.853 2.27! 44.4 <20 Years from Onset 12.683 89,462 325 36.3 20 + Years from Onset 12,051 77,391 1,946 53.8 - occurred '. for Dur;r : :8.9c U.S. Nj result oi the deat'. number . Del -- Unde Total, Total , Car. Piet pcr \tc. Car Car Car Car Car Car Al! Nor.ir d As" A!! o: *Expc National tRjic (BE) (autops1(DC) and ! 15 (esopha Car expcctc oro-ph^ twice : interop Addin* Res betwee Selikoff et al.: Mortality of Insulation Workers 103 occurred 1946 deaths among 12.051 men who had achieved 20- years from onset. 1. 1967, with 77,391 man-years of observation. The average age during observation was 36.3 >s Dust years for the shorter group and 53.8 years for those with longer duration from onset. =g= During the decade of observation 2271 deaths occurred (Table 12), whereas only ' 1658.9 deaths were expected (based upon white male age specific mortality data of the :os i U.S. National Center for Health Statistics). The excess deaths were primarily the 40.^9 result of an increased number of instances of cancer of several sites. Investigation of -------------------- the deaths found 486 due to bronchogenic carcinoma, between 4 and 5 times the ' number anticipated. There were 175 deaths of mesothelioma, 63 pleural in iocation 72 179 201 21 105 71 37 50 16 31 7 20 617 193 )n of Heat and ned. Most of t categorized in the tables n onlv (DC). yf 40(10.101 sure (12.683 anged its age the men who s from onset with 89,462 years. There o States t Years m Onset 2.051 77.391 1.946 53.8 Table 12 Deaths Among 17,800 asbestos Insulation Workers in the United States and Canada January 1. 1967-D6CEMBER 31, 1976 Number of Men 17,800 Man-years of Observation 166.853 Underlying Cause of Death Total deaths, all causes Total cancer, all sites Cancer of lung Pleural mesothelioma Peritoneal mesothelioma Mesothelioma, n.o.s. Cancer of esophagus Cancer of stomach Cancer of colon-rectum Cancer of larvnx Cancer of pharvnx. buccal Cancer of kidnev All other cancer Noninfectious pulmonary diseases, total Asbestosis All other causes Expected* 1658.9 319.7 105.6 t t + 7.1 * 14.2 38.1 4.7 10.1 8.1 131.8 59.0 + 1280.2 Observed (BE) (DC) 2271 995 486 63 112 0 18 7*3 59 11 21 19 184 2271 922 429 25 24 55 18 18 58 9 16 18 252 212 168 1064 188 78 1161 Ratio o/e (BE) (DC) 1.37 3.11 4.60 -- -- -- 2.53 1.54 1.55 2.34 2.08 2.36 1.40 1.37 2.88 4.06 -- -- -- 2.53 1.26 1.52 1.91 1.59 2.23 1.91 3.59 3.19 ---- 0.83 0.91 Expected deaths are based upon white male age-specific U.S. death rates of the U.S. National Center for Health Statistics, 1967-1976. tRates are not available, but these have been rare causes ofdeath in the genera! population. (BE): Best evidence. Number of deaths categorized after review of best available information (autopsy, surgical, clinical). - (DC): Number of deaths as recorded from death certificate information only. and 112 peritoneal. There was a modest increase in deaths of gastrointestinal cancer (esophagus, stomach, coion-rectum). Cancers of several other sites were also significantly increased over the number expected. There were 11 deaths of cancer of the larynx. 21 of the buccal cavity and oro-pharynx, as well as 19 deaths of cancer of the kidney. In each instance, this was twice the number anticipated. Comment had previously been made: concerning interest in these sites but. at the time, there were inadequate data for evaluation. Additional experience has indicated that the early hints were predictive. Review of information concerning the deaths allowed a distinction to be made between deaths caused by asbestosis and those of other noninfectious pulmonary 104 Annals New York Academy of Sciences diseases, including emphysema and cor pulmonale. Thereby. 168 deaths were assigned to asbestosis and 44 were attributed to other noninfectious pulmonary diseases, compared with a total expectation of 59.0 in the general population. The question of whether noninfectious pulmonary diseases, other than asbestosis. are or are not increased in asbestos workers is a complex problem. Continuing questions of nomen clature of "chronic obstructive lung disease" in general, and uncertainties of designa tion and pathological categorization were recently well reviewed by C. M. Fletcher.7 Chronic nonspecific lung disease, particularly that due to cigarette smoking, may have an influence on the course of the asbestosis, providing an added burden to damaged lungs. We have evidence that there is such influence.5 On the other hand, diffuse interstitial parenchymal fibrosis or diffuse pleural fibrosis would be disadvan tageous for individuals with extensive chronic obstructive lung disease, particularly with superimposed pulmonary infections or other physiological stress. Such disadvan tageous additional contributions, in individual cases, may play important roles in determining the adequacy of a patient's respiratory reserve. Categorizing deaths in terms of a single underlying cause, however, does not allow for inclusion of nuances of interaction among the several factors, nor to depict the full spectrum of such interacting disease, from cases in which little other than asbestotic pulmonary fibrosis may be involved to those in which all evidence indicates that cigarette smoking and chronic obstructive lung disease were the principal factors. For causes other than cancer and asbestosis and other noninfectious diseases, there were fewer deaths than expected, 1064 against 1280.2. In particular, there were fewer deaths of arteriosclerotic cardiovascular disease and its consequences. This is further considered elsewhere,* and it may well be that the active physical work associated with employment as insulation workers is advantageous in preventing or delaying death of cardiovascular disease, although the influence of initial selection upon admission to the trade cannot be disregarded. It is evident that most excess deaths in this cohort were due to cancer (675) rather than to asbestosis and other noninfectious pulmonary diseases (153). This is notewor thy, since the present United States Standard for occupational exposure to asbestos is in large part derived from reported recommendations designed to prevent asbestosis.' Deaths of Less Common Malignant Neoplasms Apart from lung cancer, mesothelioma, gastrointestinal cancer, cancer of the larynx, pharynx and oral cavity and cancer of the kidney, there was still as excess of cancer of other sites, with 184 observed, compared with 131.8 expected. Table 13 provides information concerning expected and observed deaths for a number of other sites. For some, there was no evidence of increased incidence, as with leukemia, lymphoma, primary cancer of the liver, testes, bladder. Relatively small increases were suggested for brain, skin, pancreas, prostate. The overall increase is of some interest, especially in view of the known possibility of asbestos fibrils being dissemi nated to virtually all organs following inhalation or ingestion.10" We have long been properly accustomed to seeking verification of increases beyond chance expectation ("statistically significant") in incidence of cancers of defined sites. This perspective does not contradict the potential importance of more general increases in cancer incidence of a large variety of sites, perhaps not reaching levels of statistical significance in any one location but still yielding definite increases of the overall cancer burden of the groups investigated. Fron of sever; of these of such would n is the us neoplasr neoplast conditio among : Dev UndTotal Canoe Death r Par L;v Bio ToPrc Le; Lvr Sk: Br- Expc N3tior.a (BE) (autc~-v (DC i (Tabl:these t howeve and th. death. 10 oth "incido experic proper cancer assigned diseases. :estion of are not f nomen* ' designaetcher.7 :ing, may urden to ler hand, lisadvanticuiarly iisadvanroles in deaths in jances of of such y fibrosis king and ses, there ere fewer s further ited with death of fission to 5) rather noteworibestos is estosis.9 r of the excess of ABLE 13 of other :ukemia, ncreases of some dissenti ng been ectation spective ; cancer atistical : overall Seiikoff el al.\ Mortality of Insulation Workers 105 Multiple Cancers From a purely statistical point of view, in view of the increased incidence of cancer of several sites among asbestos insulation workers, we would expect that a proportion of these men would suffer multiple cancers simultaneously, even beyond the tendency of such findings to be made among individuals with cancer, in general.12 Again, this would not be reflected in tabulations of causes of death by single underlying cause, as is the usual practice. Analysis of our experience demonstrated one hundred malignant neoplasms present but not causing death (Table 14). Sometimes these additional neoplasms were mentioned on the death certificate but as an "other significant condition," not in the section on the underlying cause of death. Forty were pre *.t among the 1064 cases where death was due to causes other than cancer or asbestosis Table 13 Deaths Among 17.SOO asbestos Insulation Workers in the United States and Canada January 1. 1967-December 31. 1976 Number of Men 17,800 Man*years of Observation 166.833 Underlying Cause of Death Total deaths, all causes Cancer, ail sites Deaths of less common malignant neoplasms Pancreas Liver, biliary passages Bladder Testes Prostate Leukemia Lymphoma Skin Brain Expected* 1658.9 319.7 17.5 7.2 9.1 1.9 20.4 13.1 20.1 6.6 10.4 Observed (BE) (DC) 2271 995 2271 922 Ra tio o.'e (BE) (DC) 1.37 1.37 3.11 2.88 23 49 1.32 2.81 5 19 0.70 2.65 9 7 0.99 0." 2 1-- -- 30 28 1.47 1.37 15 15 1.15 1.15 19 16 0.95 0.30 12 8 1.82 1.22 14 17 1.35 1.63 'Expected deaths are based upon white male age-specific U.S. death rates of the U.S. National Center for Health Statistics, 1967-1976. (BE): Best evidence. Number of deaths categorized after review of best available information (autopsy, surgical, clinical). (DC): Number of deaths as recorded from death certificate information only, (Table 15). Among the 168 deaths of asbestosis. cancer was aiso present in 7, 6 of these being bronchogenic carcinoma. Analysis of the circumstances leading to death, however, indicated that the underlying cause was asbestotic pulmonary insufficiency, and that the lung cancers were present but with no decisive influence at the time of death. Nineteen other cancers were present among the 486 deaths of lung cancer and 10 other cancers accompanied the 175 deaths of mesothelioma. There were 9 "incidental" neoplasms among the 99 deaths of gastrointestinal cancer: Although experiences are so far limited, it may not be wholly unexpected that there were proportionately more incidental neoplasms accompanying deaths of colon-rectum cancer, compared to those of lung cancer (8.5% vs. 3.9%). One may speculate that this 106 Annals New York Academy of Sciences Table 14 Mortality Experience Among 17,800 Asbestos Insulation Workers in the United States and Canada 1967-1976: Observations in 2271 Consecutive Deaths Malignant Neoplasms Present, but not Causing Death* Site Number Lung Pleura! mesothelioma Peritoneal mesothelioma Esophagus Stomach Colon Oropharynx Larynx Kidnev Other 24 2 l '0 1 19 3 5 3 42+ ioo; Twenty-one of these neoplasms were mentioned on the death certificate (but were not categorized as underlying cause of death), tlncluding leukemia 5, lymphoma 3, bladder 5. prostate 13, thyroid, etc. tin 92 individuals; total includes multiple cancers in eight cases. could be due to the longer clinical course of many patients with colon-rectum cancer, compared to lung cancer, with greater opportunity, simply in terms of time, to develop additional disease. Multiple cancers were present, overall, in 2.1% of deaths among these asbestos insulation workers (48 of 2271). It is perhaps to be expected that this was more likely to be the case among those for whom cancer was the primary cause of death (4.5%) while only 3 of the 1276 other deaths had this finding. From a clinical point of view, in the management of patients with asbestosassociated disease, the potential for incidental or multipie cancers is of some importance, and awareness of the possibility may assist in both the diagnostic investigation and long-term surveillance of individuals with history of significant asbestos exposure. Lapsed Period It is now well appreciated that most asbestos associated disease is first seen after considerable periods from onset of exposure in both occupational and environmental circumstances. This is true both for the presence and extent of parenchymal fibrosis and pleural fibrosis and/or calcification,JJJ and for asbestos-associated neoplasms.14 Data obtained in this investigation now provide a more complete overview of this question. Broadly, we saw some limited excess disease in less than 20 years from onset of exposure (Table 16). Among 12.683 men with such experience, covering 89,462 man-years of observation, the number of cancer deaths was about doubled, with 42.6 deaths expected and 83 observed. There were no excess deaths of gastrointestinal cancer and only 5 deaths of mesothelioma, with these in the 15-19 years from onset category. Age, year and sex specific mortality data of the U.S. National Cancer for Health Statistics indicated that 11.9 deaths of lung cancer were to be expected. Thirty-six occurred. There were 8 deaths of asbestosis. On the reached 2C were antic cancer. It tion, with deaths of found. Tai employme death. Th cannot be number o: number c;' not take ir achieved c in number with conn lioma. bot reaching i 35-39 ye; thousand ; In am indicates t years frorr Mortalit STAT: Number to _____Ur; Cancer Cane Pleur Peri'. Cane Cane Cane Cane Cane Cane AH a Noninf: d:> Asbe All oihc. Tou Six of  United \THS c were not im cancer, to develop e asbestos nore likely ath (4.5%) i asbestosi of some diagnostic significant . seen after ironmental nal fibrosis olasms.14 iew of this from onset ing 89,462 :. with 42.6 rointestinal from onset Cancer for ; expected. SeiikofF et alMortality of Insulation Workers 10' On the other hand, extensive disease was seen among the 12.051 men who had reached 20 or more years from onset during the decade .of study. Here, 1376.0 deaths were anticipated; 1946 occurred. There were 160 deaths of asbestosis and 912 of cancer. It was at this time that bronchogenic carcinoma made its heaviest contribu tion, with 93.7 such deaths expected and 450 observed. One hundred and seventy deaths of mesothelioma were then seen and the increase in gastrointestinal cancer found. Table 17 depicts these data in some detail, in five-year periods from onset of employment. Lung cancer data are given as both expected and observed numbers of death. This practice cannot be followed for mesothelioma, where expected deaths cannot be computed for the general population. Instead, we have provided data in both number of deaths of pleural and peritoneal mesothelioma, as well as in terms of number of deaths of these causes per thousand persons years at risk. The latter does not take into account variations in achieved age. but this may have less influence than achieved duration from onset of employment. It will be seen that very major increases in numbers of deaths of lung cancer are first seen at 15-24 years from onset of work, with continued further increases. The extraordinary increase in deaths of mesothe lioma, both of the pleura and the peritoneum, is not observed until somewhat later, reaching 2.78 deaths per thousand person-years at risk for pleural mesothelioma at 35-39 years from onset of work, and 5.47 deaths of peritoneal mesothelioma per thousand person-years at 45+ years from onset. In another reflection of the clinical concerns among these workers. Table 18 indicates that approximately one-third of all deaths were due to lung cancer at 30-34 years from onset, while mesothelioma accounted for 13% of all deaths at 35-39 years. Table 15 Mortality Experience Among 17,800 asbestos Insulation Workers in the United States and Canada 1967-1976: Observations in 2271 Consecutive Deaths Number of Incidental Malignant Neoplasms (not Causing Death) in Relation to Underlying Cause of Death as Established by Best Evidence (BE) Underlying Cause of Death Cancer all sites Cancer oflung Pleural mesothelioma Peritoneal mesothelioma _ Cancer ofesophagus Cancer of stomach Cancer ofcolon-rectum Cancer of larynx Canccrof pharynx, buccal cavity Cancer of kidney Al! olher cancers Noninfectious pulmonary diseases, total Asbestosis All other causes Total Number of Deaths of Underlying Cause 995 486 63 112 18 22 59 11 21 19 184 212 168 1064 2271 Incidental Malignant Neoplasms No. of Deaths Total Cancers 45 50 17 19 44 56 11 33 55 00 23 00 89 10 10* 7 7' 37 40 92 , 100 `Six of these were lung cancer 108 ::&\i -**r* Annals New York Aeademv of Sciences Tahi v. 16 Deaths Among 17,800 Asbestos iNfjiii ation Workers in the IJniteo States and Canada January I, 1967-December 31, 1976, Anai Ysis by Duration i rom Onset <>i Fmpioyment Total Men Man-year** of Observation Underlying Cause of Dealh Total deaths, all causes Cancer, all sites Cancer of lung i Pleural mesothelioma Peritoneal mesothelioma Mesothelioma, n o s. Cancer of esophagus Cancer of stomach Cancer of colon-rectum C ancer of larynx Cancer of pharynx, buccal Cancer of kidney All other cancer Noninfectious pulmonary diseases, total Asbcstosis All other causes . Before 20 Years from Onset 12,683 89,462 Observed Ratio o/e F.xpected* (HI) (lt) (DC) 282.9 42 6 1 1.9 t t t 06 1.5 41 04 u II 21.7 5.2 t 215.1 36 12 3 03 2.69 22 i 0-- -- 0 1-- -- 1 1---- i 0-- -- 4 4-- -- 2 2- -- 1 2-- -- 33 2X 30 1.29 I.1K 8 11 1.54 2.12 X 2-- -- 214 2.17 1.00 1 01 20 or More Years from Onset 12,051 77,391 Observed Ratio o/e F.xpected* (BIO (DC) (HI*) (DC) 1376.0 277.1 91.7 t t t 65 12 7 34 0 4.3 8.8 7.0 not 51.X t 1 045.1 194 n 912 450 61 109 0 17 21 55 9 IK 16 156 204 160 810 1946 84 5 397 21 24 54 17 IK 54 7 14 15 222 177 76 924 4 80 -- -- 2 64 1 65 1.62 2.09 205 2.29 1.42 3.78 -- 0.79 4.24 -- -- 2 64 1.42 1 59 161 1.59 2.14 2.02 3.28 -- 0 88 Expected deaths arc based upon while male age-specific 1 I S dealh rales of the U S National Center for 1 leallh Statistics, 1967 1976. tKaies arc not available, bul these have been rare causes of death in the general population. (HP). Best evidence Number of deaths categorized after review of hcsl available information (autopsy, surgical, clinical). (DC): Number of deaths as recorded from death certificate information only. tRales are noi uvaii.iuii., uni iiic.so luv. l*.ch ..i.c vuu .^ ...... ............... .. k ,, (BE): Best evidence. Number of deaths categorized after review nf best available information (autopsy, surgical, clinical). (DC): Number of deaths as recorded from death certificate information only. Selikoff et al.\ Mortality of Insulation Workers Tahie 17 Deaths Among 17,800 Asbestos Insulation Workers in United States and Canada. January l, 1967 I)e<'emueh 31, 1976. Analysis y /Juration i-kom Onset oe Emeioyment Lung Cancer Pleural Mesothelioma Peritoneal Mesothelioma Duration from Onset (Years) <10 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45 + Number of Men H.190 9.063 9.948 K.8K7 6.5tj6 3,547 2.020 1,108 1,448 Person-years of Observation 26.393 29.003 34.066 31.268 20.657 11.598 5.403 3,160 5.305 Exp.* 0.7 2.7 8.5 17.0 21.0 18.4 11.5 8.1 17.8 (Bli) O 7 29 59 105 112 65 40 69 (DO 0 5 27 57 96 103 57 131 53 ___ --___ (BE) (DC) -- -- 2.55 1.82 3 40 .3.17 3.48 3.36 5.00 4.58 6.08 5.59 5.68 4.98 4.91 3.82 3.89 2.98 ------ . ---- -- (HE) (DC) 00 00 22 64 LI 5 93 15 4 4J 14 4 No./1000 Person- (BE) 0 0 0.06 0.19 0.61 0.78 2.78 .1.27 2.64 Number (BE) 0 0 3 3 19 23 19 16 29 (DC) 0 0 0 2 3 6 .5 3 5 No./1000 Person- (BE) 0 0 0.09 0.10 0.92 1.98 .1.52 5.06 5.47 Expected deaths arc based upon white male agc-spccilic U S. dentil rates of the U.S. National Center for Health Stulistics, 1967 1976: Smoking habits not taken into account. (BE): Best evidence. Number of deaths categorized after review of hcsl available information (autopsy, surgical, clinical). (DC): Number of deaths as recorded from death certificate information only. *>s*4f.sr. Annals New York Academv of Sciences Table 18 Deaths Among 17.800 Asbestos Insulation Workers in the United States and Canada, January 1, 1967-December 31, 1976. Analysis by Duration from Onset of Employment Years from Onset of Employment <10 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45* Total Total Deaths 51 85 189 320 388 340 253 203 442 2271 Lung Cancer (BE) (DC) 00 8.2 5.9 15.3 14.3 18.4 17.8 27.1 24.7 32.9 30.3 25.7 22.5 19.7 15.3 15.6 12.0 21.4 18.9 Percent of All Deaths Mesothelioma Pleurai (BE) (DC) 00 00 1.1 1.1 1.9 1.3 3.4 1.3 2.7 0.9 5.9 1.6 2.0 1.5 3.2 0.9 2.8 1.1 Peritoneal (BE) (DC) 00 00 1.6 0 0.9 0.6 4.9 0.8 6.8 1.8 7.5 2.0 7.9 1.5 6.6 1.1 4.9 l.l Total (BE) (DC) 00 00 2.7 1.6 2.8 2.5 8.3 s > 9.4 6.5 13.4 7.9 9.9 6.4 9.7 4.1 7.7 4.6 Total includes mesothelioma not specified as either pleural or peritoneal. (BE): Best evidence. Number of death categorized after review of best available information (autopsy, surgical, clinical). (DC): Number of deaths as recorded from death certificate information only. Altogether, lung cancer was responsible for 21% of all deaths observed by us in this cohort and mesothelioma for 8%. Investigation and Categorization of Causes of Death It is widely known that causes of death as recorded on death certificates may be in error and that, even when accurate, these may be coded with considerable variation among different agencies, despite the existence of agreed upon international rules and recommentations.15 Yet death rates based upon large series of causes of death as recorded on death certificates are nevertheless useful, and are widely utilized. This distinction, however, leads to the understanding that comparisons between observed causes of death and those expected from national or local "death rates" should not be considered in rigid exact terms, especially when considering the mortality experience of very specific groups. For these, there are no perfect "controls" that would take into account their age distribution, ethnic derivation, smoking habits, economic circum stances, prior personal and social history, and so on. ` The matter is further complicated by the fact that deaths in the general population, which provide the basis for established comparison death rates, are not verified by investigation or examination of available data concerning the circum stances associated with the deaths. Of course, one may elect to treat the deaths in the study population in exactly the same way, with no attempt to review data other than that recorded on the death certificate, explicitly accepting whatever errors might exist in causes of death recorded and implicitly hoping that whatever errors exist are very much the same in the two sets of data, deaths in the general population and those in the group under investigation. Certainly this latter method is simpler, cheaper, much less time consuming.16 v Bt cause: popui; differ: death report this d: only these and e: been : perito revisit Ju is uni availa at aui Lr. Cane Ca p: Pc M Ca Ca Ca Ca Ca Ca Ca Ca Ca Ca Nom: c Ay All o'. Center tR.n (BF; (DC S AND (DC) 0 0 u 2.J 5.2 6.5 7.9 6.4 4.1 4.6 armation s in this ay be in ariation ules and :eath as ed. This observed d not be oerience ake into circum- general are not circumts in the her than ght exist those in r, much SelikofT et a!.: Mortality of Insulation Workers Hi But this approach has a number of drawbacks. First, where the distribution of causes of death in the group being studied is different than that found in the general population, one may expect that the distribution of inherent error might also be different. Second, data may be available which can shed much light on the causes of death under investigation and not including such information permits the risk of reporting results which are simply inaccurate. It is perhaps inadequate comfort to skirt this difficulty by clearly stating that only death certificate diagnoses are reported; this only gives the reason for potential inaccuracy. There are also technical issues, and these are of particular concern among asbestos-exposed groups, in the sixth, seventh and eighth revisions of the International Classification of causes of death. There have been no clear categories for the diffuse malignant mesotheliomas of the pieura and peritoneum which occur, and while some improvements have been made in the ninth revision now being introduced, we fear that these will still not resolve the problem. Just as there are no perfect controls, there is also no single perfect solution, and it is unlikely that there will be such in the foreseeable future, especially since data available for ascertainment of cause of death, even when investigations during life and at autopsy have been undertaken, are not always complete and are surely not uniform. Table 19 Mortality Experience Among 17,800 asbestos Insulation Workers in the United States and Canada 1967-1977: Observations in 2271 Consecutive Deaths with Underlying Cause of Death Coded according to Death Certificate Information Only and according to Best Evidence Death Certificate (DC) Best Evidence (BE) of Underlying Cause of Death Expected* No. o/e No. o/e Evidence Cancer all sites Cancer of lung Pleural mesothelioma Peritoneal mesothelioma Mesothelioma, n.o.s. Cancer of esophagus Cancer of stomach Cancer of colon-rectum ' Cancer 61 larynx, pharynx, buccal cavity Cancer of kidney Cancer of prostate Cancer of bladder Cancer of pancreas Cancer of liver Cancer of brain Noninfectious pulmonary diseases, total Asbestosis All other causes 319.7 105.6 t t t 7.1 14,2 38.1 14.1 8.1 20.4 9.1 17.5 7.2 10.4 59.0 t 1280.2 922 2.88 429 4.06 25 -- 24 -- 55 -- 18 2.53 18 1.26 58 1.52 995 3.11 486 4.60 63 112 -- -- 0-- 18 2.53 22 1.54 59 1.55 25 1.69 18 2.23 28 1.37 7 0.77 49 2.81 19 2.65 17 1.63 32 2.16 19 2.36 30 1.47 9 0.9923 1.32 5 0.70 14 1.35 188 3.19 78 -- 1161 0.91 212 168 1064 3--.59 0.83 93 88 40 2! -- 100 82 98 78 95 93 78 213 380 121 89 46 109 Expected deaths based on while male age-specific U.S. death rates of the U.S. National Center for Health Statistics. 1967-1976. tRatcs not available, but these have been rare causes of death in the general population. (BE): Best evidence. Number of deaths categorized after review of best available information. (DC): Number of deaths as recorded from death certificate information only. I r 1 i i ws.Y-- v'Y>- '* - ,, *?'' ' * -Jr Annals New York Academv of Sciei We have elected to approach this problem by providing dai including causes of death as recorded on the death certincate m uu, ___ comparison with "expected" deaths based upon the data of the U.S. National Center for Health Statistics, as weil as causes of death established after detailed review of all available information. Such data were retrieved in 1961 of the 2271 deaths. However, detailed information was not equally available for ail categories of causes of death. This was to be expected, in the nature of things. As a rule, the best available information for establishing the cause of death was considered to be autopsy findings, with pathological information derived from surgical intervention next, and. in their absence, clinical and roentgenological observations made during life, particularly in the period before death. Where no such details were available, the cause of death as recorded from death certificate information was then utilized. Among the 995 cases of cancer, we were required to depend upon the death certificate alone in only 28.~~wlth ofher iiifurirttilion available in 967 cases (97%). In all 175 cases of mesothelioma, surgical and/or autopsy findings were utilized. We depended upon death certificate information alone in 10 of the 486 cases of lung cancer. Similarly, in 166 of 168 deaths of asbestosis. information was available in addition to the death certificate, leaving dependence of the latter in only 15c. In contrast, for causes other than cancer and asbestosis, additional information was available in only three-quarters of the cases. Analysis of the distribution of deaths within this category showed that this was by no means unexpected. Sudden death attributed to myocardial infarction or cerebrovascular accident is often not further investigated nor need clinical abnormalities have preexisted. It was therefore no surprise that in 156 cases the medical attendant or the medical examiner warranted such death certificate diagnosis without other information being available. It would seem difficult to avoid differences in quality of ascertainment, under such varied circumstances. Analysis of our data suggests that, by and large, death certificate diagnoses were not far off the mark. This speaks highly for the diagnostic acumen and quality of care provided to these men by their medical attendants. Cancer as the underlying cause of death was indicated by the death certincate in 922 instances compared with 955 so categorized after considering all available information (Table 19). The specific diagnosis <-,f )ung and cancer of the several sites of gastrointestinal cancer, as well as cancer of the larvnx. OTO-pharvnx and kidnev was also generally well predicted by the death certificate diagnosis. in a numoer oi instances, however, there was important disparity between causes of death as recorded on the death certificate and those judged to be the case after review of available clinical and pathological material. This was true for pleural and peritoneal mesothelioma, asbestosis, cancer of the pancreas, primary cancer of the liver, cancer of the brain and chronic obstructive lung disease. In 49 cases, cancer of the pancreas was listed on the death certificate as cause of death. Only 17.5 such deaths were expected (Table 14). If we were to accept cause of death as listed on the death certificate to establish the '"observed" number of deaths for this disease, we would have to conclude that cancer of the pancreas is significantly increased as a cause of death among asbestos insulation workers. But this is not the case; when all available material was reviewed, it was found that only 23 deaths compared with the 49 so categorized on the death certificate were due to cancer of the pancreas. Four were the result of metastatic lung cancer, 15 were found on review of histological material to be cases of peritoneal mesothelioma and 5 were best categor ized as abdominal carcinoma, primary site not established. Two were due to cancer of the colon. This is not to say that we have established that there is no increased incidence of cancer of the pancreas among asbestos workers. It is possible there is some limited increase and wc are cognizant of the uncertainties inherent in the m/rr. per no: k:. be i6 r-: tor tin Us', me as; re; pc cat ica the So dc. be; COl oft W ex: ad on ro>. we m.; ma ca en so: me re'(c: s:> F-ca ur (: e than one way, i our cases, for National Center led review of all eaths. However, :auses of death. : best available .utopsy findings, xt, and, in their , particularly in ause of death as the 995 cases of in only 28, with f mesothelioma, Jeath certificate lable in addition In contrast, for ivailabie in only tin this category :d to myocardial gated nor need tat in 156 cases leath certificate iifficult to avoid es. diagnoses were i quality of care erlying cause of red with 955 so 0. The specific gastrointestinal o generally weil between causes ; the case after for pleural and y cancer of the rate as cause of accept cause of mber of deaths is significantly : this is not the only 23 deaths to cancer of the nd on review of re best categorlue to cancer of is no increased >ossiblc there is nhercnt in the Selikoff et al.\ Mortality of Insulation Workers 115 computation of "expected" rates for cancer of the pancreas since it may well be that some of the cases so categorized in the general population from which the expected rates were derived, might really be instances of lung cancer, colon cancer and even, perhaps, peritoneal mesothelioma. Since these deaths in the general population have not been investigated to verify death certificate cause of death, it is not possible to know whether such inaccuracy exists, nor its degree. Asbestosis was another example where categorization by death certificate would be misleading since only 78 deaths of this disease were so categorized compared with 168 after review. A variety of other diagnoses were offered in the discrepant cases, primarily those of "chronic obstructive lung disease." or other noninfectious respira tory disease, acute infections, and, in two cases, cancer of the lung. It was of interest that in 16 of the 97 cases called asbestosis after review but in which asbestosis was not listed as the underlying cause of death, it was mentioned on the death certificate under the rubric of "ocher significant conditions." In 129 other cases, asbestosis was again mentioned on the death certificate, but not as underlying cause of death. Altogether, asbestosis was mentioned in 230 of the 2271 death certificates. In 7 cases, asbestosis so categorized by death certificate designation was reassigned to other categories after review, with death found due to lung cancer in 5, peritoneal mesothelioma in one and myocardiai infarction in another. Accuracy of diagnosis of mesothelioma, and its being recorded on death certifi cates. is a special problem. First, there are the difficulties and subtleties of patholog ical diagnosis, of particular importance when cases of this neoplasm are seen throughout the United States and Canada by pathologists with varied experience. Sometimes, there is the added difficulty of extending the pathological diagnosis to the death certificate, not infrequently completed by a physician who may or may not have been the regular medical attendant and fully cognizant of pathological findings: to this could be added the problem occasioned by the fact that the death certificate must often be completed before there has been full consideration of postmortem findings. When these differ with the death certificate diagnosis, an amended certificate is expected to be filed. In our experience, this is rarely done. When to these are added the administrative uncertainties and insecurities of ciassificaticn. it is not surprising that only about one-quarter of the deaths of mesothelioma in this series were correctly recorded on death certificates as pleural or peritoneal, although 104 of the 175 cases were recorded as "mesothelioma," even if not with full details. We are presently analyzing the 175 deaths of pleural and peritoneal mesothelio ma. and the results will be reported. In 108 of the 112 cases of peritoneal mesothelio ma, surgical and pathological material was submitted to us for review and this was the case as well as 61 of the 63 pleural mesotheliomas (96.5% and 96.8%). The association of pleural and peritoneal mesothelioma with prior asbestos exposure is so striking" that it is not surprising that overdiagnosis might now sometimes occur. We found such to be the case in this study. In 8 cases in which mesothelioma was recorded as underlying cause of death on the death certificate, review of the histological material showed the diagnosis to be other than mesothelioma (cancer of the lung in 5 cases, metastatic carcinoma in l and Wegener's granulomato sis involving the pleura in another). In the eighth case, pleural mesothelioma was present but was not the cause of death, which was better attributed to concurrent cancer of the stomach. Where mesothelioma appeared as the diagnosis on the death certificate, depending upon the exact phrasing used, this might be coded in the eighth revision of the International List in categories 158.9 (malignant neoplasm of peritoneum), 163.0 (malignant neoplasm of pleura). 197.0 (secondary malignant neoplasm of lung). 195.0 (malignant neoplasm of abdomen). 198.9 (secondary malignant neoplasm, specified as -If-* 114 Annals New York Academy of Sciences secondary), 199.0 (multiple malignant neoplasm [carcinomatosis, disseminated cancer]), 199.1 (malignant neoplasm without specification of site) and, not infre quently, 228 (benign neoplasm of other and unspecified organs and tissues). There is at the moment no reliable way to extract the numbers of pleura) and peritoneal mesotheliomas from death certificate data categorized and recorded by health statistics agencies in the past, using the several revisions of the International Classification of causes of death. We might add that, overall, pathologists' diagnoses were much more accurate than death certificate characterization of causes of death would lead one to believe. This again bespeaks a high degree of competence on the part of pathologists in the United States and Canada (we are aware, of course, that diagnostic suspicion may be increased in those instances in which it is known that asbestos exposure had occurred, as with individuals among whom occupational history indicated such exposure; this could perhaps have accounted for some of the over-diagnosis). However this might be, it is clear that a problem exists in translating the information obtained in pathological study to the death certificate diagnosis, as well as a problem of subsequent coding of the stated cause of death. _ Conclusions Asbestos insulation workers in the United States and Canada suffer an extraordi nary increased risk of death of cancer and asbestosis, associated with their employ ment. This includes increases in death of lung cancer, pleural mesothelioma, perito neal mesothelioma, cancer of the esophagus, colon and rectum, cancer of the larynx, oro-pharvnx, kidney and perhaps stomac^TSome'Tncreases were seen'm cancer of several other sites, as well, but data are inadequate at this time to permit characteriza tion of their significance, although attention is called to such wider increase. Little increase in cancer deaths, nor of asbestosis, was observed in less than 15-19 years from onset of exposure. In general, the period of latency between onset of exposure and death was 2, 3. 4, or more decades. Large increases in lung cancer occurred at 15-35 years from onset whiie pleural and peritoneal mesothelioma showed their greatest incidence somewhat later. Under the conditions of exposure which obtained, the period of clinical latency for asbestosis was also prolonged and in many cases was 30-40 years or more from onset of employment. It would appear that in studies of the effects of such asbestos exposure it would be advantageous to analyze the experience of exposed individuals in duration-from-onset exposure categories, with particular reference to durations of more than 30 years. If this is not done, the risk is run of obscuring the neoplastic effects of asbestos exposure by the co-mingling of later deaths with those which may have occurred, much earlier and which would be much less likely to have been influenced by asbestos exposure. In the same way, studies should include, whenever possible, the opportunity fbr observation of the mortality experience of the individuals at risk at least 30--35 and preferably 40 or more years from onset of their exposure. Again, unless this is possible, only the very limited early effects will be identified and the full import of the exposures may not be appreciated. Our experiences demonstrate the advantages of review of all available information concerning the circumstances associated with the deaths that occur. Depending entirely upon death certificate diagnosis can lead to erroneous reports of causes of death, a problem that can be mitigated by characterizing causes of death according to the best available information. Using this approach, we found that the apparent increased incidence of cancer of the pancreas among asbestos insulation workers, indicated by death certificate diagnosis, did not really exist, and that many so ia:osis, disseminated Site) and, not infreand tissues). There is leurai and peritoneal recorded by health o: the International :h more accurate than i tne to believe. This oiogists in the United ;c suspicion may be ;posure had occurred, d such exposure; this owever this might be, tair.ed in pathological subsequent coding of a suffer an extraordi;d with their employnesotheiioma, peritocancer of the larynx. :re seen in cancer of permit characterizaer increase, ed in less than 15-19 ry between onset of rases in lung cancer r.esothelioma showed s of exposure which olonged and in many vouid appear that in .intageous to analyze 'sure categories, with ; not done, the risk is . co-mingling of later hich would be much e same way, studies ion of the mortality >ly 40 or more years he very limited early not be appreciated, .vailable information t occur. Depending reports of causes of if death according to i that the apparent insulation workers, and that many so Selikoff et ai.: Mortality of Insulation Workers designated were due to other causes. The same was found for primary cancer of the liver and cancer of the brain, both of which would have been considered of increased incidence using solely death certificate diagnosis, but were not so established following review of histological material. This, of course, couid well be anticipated with exposures to agents which increase the risk of neoplasms known to metastasize to these sites (bronchogenic carcinoma, gastrointestinal cancer). Our experiences do not indicate that the addition of amosite asbestos exposure during and after World War II to the chrysotile asbestos exposure which existed before then, and which continued, demonstrably altered the nature of the asbestos- induced diseases which resulted, nor their extent. Further, limited experiences suggest that asbestos insulation employment begun 1943-1962 carried the same type of long-term risk as that begun before that time. We have no information concerning the nature or extent of risk among insulation workers whose employment began after 1963. " ACKNOWLEDGMENTS We are grateful for the extraordinarily valuable assistance of Ms. Janet S. Kaffenburgh and Ms. Frances Perez in the tracing and acquisition of material concerning the insulation workers in these cohorts. Clearly, the completeness of the prolonged serial observations could not have been accomplished without their dedi cated and skillful help. At the same time, the leaders of the asbestos workers' union equally share our gratitude; Mr. Andrew T. Haas, and Mr. William C. Bernard. International President and Internationa! Secretary of the union. Mr. Roy J. Steinfurth, coordinator of the Insulators Health Hazard Program and the many union officers throughout the United States and Canada whose dedication and concern have permeated their work with us. Special mention should be made of officials of the New York-New Jersey metropolitan locals of the union. Mr. Jack Novak. Mr. Harry Ahrens. Mr. Terence McConnell, Mr. Jerome Market, Mr. Thomas Viscovich. Mr. William D. Fitzgerald, Mr. James Grogan, Mr. James F. Dwyer, Jr. and Mr. Howard Barnett. Our warm thanks are due. too. to the staff of the Department of Epidemiology and Statistics of the American Csnccr Society, including Mr. David A. Newman, Mr. Ashley Bodden, Mr. Edwin Stlverberg, Ms. Marie Corbo. who have with competence and dedication, assisted in the extensive processing and computation of the statistical data that were obtained. As so often nowadays, this epidemiological study could not have been completed without the integrated cooperation of many. Equally warm and strongly held gratitude is due to our colleagues. Dr. Jacob Churg, Dr. Yasunosuke Suzuki and Dr. Milton Kannerstein for their help in reviewing the many anatomical and histological specimens that were made available to us. Their skills, knowledge and competence form a qualitative bedrock for the study. Equally, we are appreciative of the generous help given to us by clinicians, pathologists, medical examiners and others in medical facilities in all parts of the United States and Canada, providing detailed information, records, roentgenograms. We obtained as much pleasure and satisfaction from our experiences with the kindliness and generos ity of medical and scientific colleagues as with the observations these enabled us to make. These acknowledgements would be sadly incomplete if they did not include our appreciation for the many at the Environmental Sciences Laboratory whose patient day-to-day help assisted the rest of us and assured the collection of the very large amount of information inherent in our data; warm thanks are extended to Selma 116 Annals New York Academy of Sciences Annenberg, Richard Ashley, Doris Fletsher, Rupert Fuller, Judith Marmor. Maria Martinez, Diane Monahan, Vera Reitman. Julia Roberts, Albert Rodriguez and Sidney Sibel. Their work was the warp and woof of the fabric of the investigation. References 1. Selikoff. I. J., J. ChuRG. & E. C. Hammond 1964. Asbestos exposure and neoplasia. JAMA 188: 22-26. 2. Selikoff. I. J., E. C. Hammond. & H. Seidman. 1973. Cancer risk of insulation workers in the United States. In: Biological Effects of Asbestos, P. Bogovski et. al., Eds.: 209-216. IARC Sc. Pub. No. 8. Lyons, France. 3. Seukoff. I. J., J. Churg. & E. C. Hammond. 1965. The occurrence of asbestosis among insulation workers in the United States. Ann. N.Y. Acad.Sci. 132/1: 112-120. 4. Frost, J., J. Georg. & P. L. Moller. 1956- Asbestosis with pleural calcification among insulation workers. Danish Med. Bull. 3(7): 202-204. 5. Nicholson, W. J. 1976. Case Study 1: Asbestos--the TLV approach. Ann. N.Y. Acad. Sci. 271: 152-169. 6. Selikoff. 1. J., E. C. Hammono, & J. Churg 1968. Asbestos exposure, smoking and neoplasia.JAMA204(2): 106-112. ' 7. Fletcher. C. M. 1978. Terminology in chronic obstructive lung diseases. J. Epidemiol. . Comm. Health 32: 282-288. 8. Hammond, E. C., I. J. Selikoff, & H. Seidman. 1979. Asbestos exposure, cigarette smoking and death rates. Ann. N.Y. Acad. Sci. This volume. 9. British Occupational Hygiene Society. Hygiene standards for chrysotile asbestos dust. 1968. Ann. Occup. Hyg. 11:47-69. 10. Selikoff. I. J. 1964. In Discussion. Ann. N.Y- Acad. Sci. 132/1: 596. 11. Auerbach. 0.. A. Conston. E. C. Hammond, L. Garfinkel & V. Parks. 1979. In press. 12. Berg. J. W. & D. Schottenfeld. 1977. Multiple primary cancers at Memorial Hospital 1949-1962. Cancer 40: 1801-1005. * 13. Seukoff. I. J. 1965. The occurrence of pleural calcification among asbestos insulation workers. Ann. N.Y. Acad. Sci. 132/U351-367. 14. Selikoff. I. J. Cancer risk of asbestos exposure. 1978. In: Origins of Human Cancer. H. H. Hiatt. J. D. Watson, and J. A. Winston. Eds.: 1965-1984. Cold Spring Harbor New York. 15. Percy. C. & A. Dolman. 1978. Comparison of the coding of death certificates related to cancer in seven countries. Publ. Hlth Reports. 93:(4): 335-350. 16. Enterline. P., P. De Couple. & V. Henderson. 1973. Respiratory cancer in relation to occupational exposures among retired asbestos workers. Brit. j. Industr. Med. 30: 162 166. _ 17. Cochrane. J. C. & I. Webster 1978. Mesothelioma in relation to asbestos fibre exposure. S. A. Med. J. 54: 279-281. erou com dur: par. incwiih reco cum rr.or l infer re pc Soc:. 19"C wer'sto b< 10": com 1, l in pnea: 194: disc, prev C num alloc T year: age ( men Obsc on th ages meth num-