Document MJB1maMyqdvJJZK4Zap5XKKjk

FILE NAME Firestone FIRE DATE 1977 Nov DOC FIRE034 DOCUMENT DESCRIPTION Report - Stomach Cancer Among Rubber Workers - A Control Study Ex 927 B.C. Notes - Steven Blum report Stomach Cancer Among Rubber Workers on excess rate of stomach cancer at two rubber plants Because of the common use of talc as a detackifier and the possible contamination of talc with asbestos which is a suspect stomach carcinogen the possibility exists that asbestos exposure may explain a substantial proportion of the observed stomach cancer excess 5 YS 4 ree gt sw vd ' : 7 at ymy y \ ' d NEW STOMACH CANCER AMONG RUBBER WORKERS A CONTROL STUDY Occupational Health Studies Group School of Public Health . ' University of North Carolina Chapel Hill N.C. RECEIVED LOV 2 1977 LIDUSTRIAL LIDUSTRLIIADLUSDTERIPAALRTHMYEGNTIHYEGIENE NE A Report for the National Cancer Institute Public Health Service Department of Health Education and Welfare i Pltf Exh 927 I.C. JACKSON 025435 t STOMACH CANCER AMONG RUBBER WORKERS A CONTROL STUDY by Steven B. Blum Occupational Health Studies Group School of Public Health University of North Carolina Chapel Hill N.C. A Report for the National Cancer Institute Public Health Service Department of Health Education and Welfare This project was in part supported by Grant No. 1R01 awarded by the National Cancer Institute CA 20868-01 I.C. } JACKSON 025436 TABLE OF CONTENTS Page SUMMARY we ee he ee Po a i Chapter INTRODUCTION . 2... ee eee ee eee 1 RATIONALE 2... ee ee ee ee ee ee 3 II SIGNIFICANCE 2... ee ee ee ee 9 III NATIONAL AND INTERNATIONAL RATES AND COMPARISONS 2... ee eee ee ee eee 11 IV IV VII VIII IX XI XII XIII XIV XV XVI SOCIOECONOMIC OCCUPATIONAL RELIGIOUS . AND ASSOCIATED FACTORS ......-----+-eee- 21 MIGRANT AND ETHNIC STUDIES +- + ....-. - + e+ eee DIETARY FACTORS . 2... eee ee ee ee ee eee 30 41 OTHER RISK FACTORS eee ee ee ee ee 45 ANTECEDENT CONDITIONS . eee eee a 55 PATHOLOGICAL CONSIDERATIONS ee eee --.-- 64 CARCINOGENSIS .. 2.2... ee eee ee ee eee 72 REVIEW OF MAJOR OCCUPATIONAL STUDIES ......--. 83 SUMMARY OF OCCUPATIONAL STUDY DESIGNS - MATERIALS AND METHODS ... eee- ee- ee - eee ANALYSIS I DEMOGRAPHIC AND OTHER FACTORS ...-.- 108 120 147 ANALYSIS II OCCUPATIONAL EXPOSURE DISCUSSION 2... eee ee .-...--+--.- 204 ee 256 ee JACKSON 025437 Chapter : REFERENCES . Po ee Page 276 APPENDIX I ee ee eee 286 APPENDIX II Ce ee APPENDIX III tee ee eee ee ee eee+. BIBLIOGRAPHY we ee 299 308 329 JACKSON 025438 Steven Bernard Blum Stomach Cancer Among Rubber Workers A Control Study Under the direction of Herman Tyroler Dragana Andjelkovich Carl Shy Allan Smith and Michael Symons The following report is an adaptation of the dissertation submitted to the faculty of the University of North Carolina in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Department of Epidemiology In addition to the dissertation committee identified above the following professionals contributed to this report John Gamble and Ted Williams Earl Arp - SUMMARY Mortality data collected for ten year cohorts 1964-1973 at two major tire manufacturing companies revealed an excess of deaths from stomach cancer based on elevated point estimates of Standardized Mortality Ratios SMRs at these companies The present analytic investigation was undertaken to examine possible associations between this excess and numerous occupational and occupational factors operating within the rubber industry This study utilized a matched control design with two sets of two controls for a total of four matched with each case > the first set matched on age race and sex and the second matched on these _ factors and on duration of employment as well The former control group was used to test the hypothesis of a generalized effect of employ- ment within the industry which would be evidenced by a longer duration of employment among cases The latter was used for examining for differences in the detailed work histories of cases and controls after having ensured equal opportunity by the matching constraints for each JACKSON 025439 group to have worked anywhere within the industry and these analyses alloweuds to draw conclusions concerning the risk of developing stomach w cancer in relation to work within specific areas of the plant In all there were 102 cases of stomach cancer and 393 controls with subjects from both sex and race groups included in the study - A thorough literature review of the relationship between stomach cancer and pertinent occupational and occupational risk factors is presented in Chapters III - XII of the enclosed report although these chapters need not be read for a thorough understanding of the findings After appropriate analyses the two control groups were merged and examination of the data indicated that there was no statistically significant association between stomach cancer cases and numerous potential explanatory factors such as nativity education age at hire and total duration of employment within the industry This was followed by an analysis of the detailed work history of each subject a history which contains a listing of each job held by the individual as well as the dates during which these jobs were held Using a system developed by the Occupational Health Studies Group OHSG at the University of North Carolina the thousands of unique jobs within the industry have been condensed into a more manageable number of approximately 100 Occupational Titles OTS which share common production processes and these OTs have in turn been further condensed into 20 standard Occupational Title Groups OTGs with even broader common features The work history analysis involved an examination of the histories of stomach cancer cases and their controls for different OT frequencies between the two groups a finding which would indicate possibly different JACKSON 025440 exposures to known orsuspect gastric carcinogens Initial use of the standard OTGS did not reveal differences in possible exposure After these analyses a new job classification was constructed by expert OHSG environmental hygienists This new classification grouped OTS thought to have potential for exposure to numerous known or suspect stomach carcinogens Use of this system revealed that cases at one company had significantly different work histories from controls and this difference involved a history of employment within OTs judged to have substantial exposure to detackifiers sticking agents Further analysis indicated that this exposure among the cases lasted longer and occurred in an earlier calendar period than did the control group exposures Because of the common use of talc as a detackifier and the possible contamination of talc with asbestos which is a suspect stomach carcinogen the possibility exists that asbestos exposure may explain a substantial proportion of the observed stomach cancer excess JACKSON 025441 INTRODUCTION In 1970 the United Rubber Workers Union negotiated with the six major U. S. rubber companies fora comprehensive program of research into related health problems The Occupational Health Studies Group OHSG at the University of North Carolina is conducting research within four of these companies We have access to large stable populations of industrial employees on whose health status we have already completed numerous studies with many more still in progress One of our most recent investigations has revealed a substantial excess of stomach cancer based on Standardized Mortality Ratios SMRs among rubber workers and we have accumulated evidence of the nature of the environmental agents within the industry which may be of etiologic importance The OHSG has developed methods to analyze employee work histories which is of prime importance in an industrial setting enabling us to categorize work experience into | distinct Occupational Titles OTS each of which shares common industrial processes These OTs can be further condensed into broader categories called Occupational Title Groups OTGs which are helpful in initial analyses While the OTs are not exactly similar in every company under study the OHSG has developed a classification system Appendix I which enables us to pool data across the companies and plants included in this investigation JACKSON 025442 2 Retrospective cohort studies which identified employee cohorts as of 1964 were initiated at one plant in each of two companies in Akron at which time all active and retired male workers who were alive and between the ages of 40-84 were identified The mortality experience of these cohorts was obtained from the companies life and pension insurance departments with additional data collected from various state departments of vital statistics when unavailable from company sources At both companies A and B these data were collected through 1973 so we have year cohort analyas t e bos th plants The U. S. general population was used as the referent standard from which appropriate age- and specific rates were derived for the determination of expected numbers of cases and the results are presented in Table . However since the completion of these analyses additional gastric cancer deaths beyond the termination years have been recorded at both companies CompaB nhays also collected data on the mortality experience of a running cohort for the interval 1950-63 Table 2 presents the total pool of stomach cancer deaths available for this study Despite this large number of potential inclusions the cases actually included were part of the year cohorts with the addition of cases which occurred in 1974 and the first part of 1975 The present study is based on 102 deaths from gastric cancer and examines in control mode the possible association between this disease and numerous occupational and occupational factors JACKSON 025443 RATIONALE Workers in the rubber industry are or have been exposed to a variety of chemical agents which are either clearly established carcinogens or are highly suspected of being so Table 3 presents a partial list of chemical carcinogens believed to have been present in the working environment of the rubber industry As can be seen the three classes of chemical agents benzpyrene nitrosamines and asbestos most often correlated with the incidence of gastric malignancies are either used as such or formed during the industrial process Table 2 presents the stomach cancer mortality data collected by the Occupational Health Studies Group CHSG at the University of North Carolina prior to the initiation of the present study In view of the high case fatality and the year survival rate of only 12 per cent the use of mortality instead of incidence cases seemed justified for case identification In those companies for which retrospective identification of populations at risk have been completed _ standardized mortality ratios for males are presented As is evident from Table 1 there is a 30-90 per cent higher mortality from stomach cancer for males based on comparable U. S. national mortality data It should be noted however that the indicated mortality excess in males is probably a conservative estimate since the national data are compiled for all males including those who are ridden or disabled within the specified age categories The workers comprising the populations at risk within the rubber industry however are healthy enough to be actively employed JACKSON 025444 and comparing their rates with result in an estimation national data would therefore probably as witness the SMR of less than 100 for all causes Column ) This phenomenon is commonly called the healthy worker effect ] For this reason Column 3 in Table , the adjusted- causes rate gives a closer approximation of the true increase in mortality rate In view of the evidence of increased gastric cancer mortality among rubber workers and the concomitant presence of known carcinogenic agents within the industrial environment an epidemiologic study seemed appropriate As mentioned earlier the OHSG had already collected a substantial amount of data on employees of the rubber manufacturing plants and the remainder of the pertinent information was available | A control epidemiologic study by comparing the occupational histories of cases and controls would be able to determine most importantly if the cases had some common work area or exposure not shared by would be the controls then using as If this were to be true a population only and we suspected it that group of workers employed in high risk areas would be expected to yield standardized mortality rates far higher than the overall excess reported thus far a finding which would be possible when the dilution effect is eliminated . | The control approach was chosen because it would allow us to -immediately use the ascertainments of death collected for the period ' under study In addition other typeosf information which were either immediately available or which could be feasibly collected would allow us to examine other types of variables which are suspected of being JACKSON 025445 associated with the development of cancer There are a number of advantages to this method First by analyzing both duration of employment and occupational histories we would be able to determine if length of employment within specific work areas was a good index of exposure to carcinogens and subsequent development of disease Also by gathering data on variables such as ethnicity we would be able to control for these possible confounding factors For instance mortality rates from gastric cancer show significant geographic variation Some Eastern European countries such as Czechoslovakia and Poland have mortality rates three to four times higher than those in the S. This may well have been a factor at some of the plants under study and has the potential to explain the entire observed excess within the industry The most provocative general finding has been the consistently elevated point estimates for the gastric cancer SMR in all companies and plants thus far examined indicating that any responsible agent or agents may be used extensively throughout the entire industry and not limited to only a few facilities and that the excess is probably not entirely explained by the presence of a single common confounder The control design enables us to determine if cases are generally more associated with some specific aspect of the production process both within individual plants and across the two companies than their approximately chosen controls JACKSON 025446 Table 1 Summary of Preliminary Stomach Cancer Mortality Data for Males Employed in the Rubber Industry Company A B Race Observed Number of Deaths 1 All Causes SMR 2 Stomach Ca 3 Adjusted by causes rate :( all & 39 99 187 W 34 95 124 1.89 1.31 JACKSON 025447 Table 2 Total Numbers of Stomach Cancer Deaths Ascertained from Death Certificates by Race Groups Company A 1. 1964-73 a 2 1964-73 b Race Group WM WF BM Total . 37 5 6 48 41 | 5 6 52 B 1 1964-73 a 36 4 2 42 2. 1964-73 b 41 5 2 48 7 3 1950-63 a 24 1 2 27 4 1950-63 b 28 1 2 31 a = UCD Underlying Cause of Death only b = anywhere on death certificate and not just UCD JACKSON 025448 Table 3 Partial List of Chemical Carcinogens Believed to be Present Within the Rubber Industry Nitrosamines nitrosodiphenylamine dinitrosobenzene Dinitroso dimethylterepthalamine dimethylterepthalamine dihydroquinoline Polycyclic Aromatic Hydrocarbons 3,4 benzpyrene chrysene pyrene 1,2 benzpyrene 1,12 benzperylene 1,2 benzanthracene Asbestos and Containing Talc JACKSON 025449 SIGNIFICANCE While the rate of stomach cancer mortality in the United States has been steadily declining during the last four decades rubber workers are experiencing a significantly elevated risk compared with the general population In addition the prognosis for stomach cancer is poor with a median survival time of less than five months and a year survival rate from of only 12 per animal studies cent that 2 There exists rather convincing certain environmental pollutants are evidence capable of inducing gastric malignancies and many of these agents are present in the occupational environment of the rubber industry This investiga- tion was the first analytic study concentrating specifically on the stomach cancer problem among American rubber workers The overall objective was to identify work areas or jobs in which fatal stomach cancer cases have clustered and there already was some preliminary evidence of concentration within the industry From the results of this study we hoped to be able to make definitive statements concerning the specific areas involved and we foresaw the possibility of following with detailed measurements of pollutant levels which would result either in their removal from the production process or in the development of protective devices for the workers It was also possible that this study would implicate previously unsuspected - carcinogens and thus In addition serious contribute to the consideration was existing theory of carcinogenesis also given to the possibility JACKSON 025450 10 that carcinogens which are present in the industrial environment of the rubber industry and associated with a higher incidence of gastric cancer may also be present in the ambient air in smaller doses and may therefore be affecting the general population The design of this permit - study would us to draw conclusions about the effect of length of exposure and age at first exposure on mortality from this disease Only about 10 per cent of stomach cancer diagnoses are made when the disease is in a localized stage during which surgical treatment is most effective The year survival rate for those diagnosed during the localized stage and treated surgically is 50 per cent compared with 12 per cent for all gastric cancer patients 2 The impact of identifying those specific aspects of the production process in which workers are at highest risk was thought to be considerable since workers in these areas who had already been inadvertently exposed would be alerted to their increased risk This could result in diagnoses being made an an earlier stage of the disease process thereby significantly improving the prognosis More importantly if carcinogenic agents were identified in the industrial environment they could be either removed or replaced or workers could be provided with protective gear JACKSON 025451 NATIONAL AND INTERNATIONAL RATES AND COMPARISONS Stomach cancer incidence rates vary considerably both between and within countrieansd estimates of these rates differ according to the calendar period for which they are computed In addition many researchers use mortality rather than incidence rates because the former are more easily and accurately compiled Mortality rates are considered a valid reflectioonf incidence because of the very short median survival time and the low year survival rates 2 Doll 3 Tables 4-6 and Dunham and Bailar 4 Table 7 and Figure ) have compiled data illustrating the international variation in adjusted mortality and incidence rates While there is some variation in these rates depending on which of the two are used the general rankinogf countries remains quite constant Specifically high rates were found in the USSR Chile Japan Iceland Finland Central Europe and most of Latin America Geographical areas with low rates include North America Africa and Southern and Southeast Asia Dunham and Bailar have condensed these data into a map Figure ) which illustrates the geographic areas with low intermediate and high rates They found the highest male rates based on mortality in Soviet Russia 112 per 100,000 and the lowest in Singapore 1.6 per 100,000 Doll found the highest male rates based on mortality in Japan 153.1 per 100,000 and the lowest in the United States 18.7 per 100,000 . JACKSON 025452 12 i i JapanJapan Iceland Iceland Iceland Chito Numeli Cancer of the the stomach i , t ! Country Country Country Country . ry toe . noe oo s EnglandEngland UgendaMosambiquMoesambique from 3 TABLE 4 perper per 100,0010 00,000 100,000 100,000 patsonpsatsons patsons patsons patsons patsons 15 1651 6115 61 10,0 10,010,0 10,0 Rato 10,0010,00 . i i Rato per per 100,000 | | 1 t i : Rato j i i t [| |! . 6 . 221. 20 138.0 1 20 . 104.3 69 4.6 2. 69 104.3+ ? 5201..0 3 . 50.0 50.0 F . i H ; !i |; | ; ' !| \ 1 | 4646 Ho:T 44.6 44.6 44.6 47 27 42. 1.9 2. 5 2.2. 19 19.519.5 45.2 . 44.7 . 4943. 2.2 2.2 1.9 ~ 2. 1 2. 2.3 24 410 . 32.7 , 2.7 2.7 1 147 29.7 . 26Aen 147 . OD 214 214 24 24 2020 10 23.2 9.6 10 23.2 23.2 8.7 . 2 1 ed2 6 1 JACKSON 025453 13 TABLE 5 Cancer of the stomach annual standardized 45-04 years old standardized mortality rates per 100,000 persons age ; Country Rate per 100,000 M F Male rate : unit | female rate : 153.1 78.3 2.0 144.8 81.6 1.8 100.0 47.2 2.1 , 05.7 41.3 .2.3 80.7 36.6 2.4 85.9 37.S 2.3 84.1 70.8 42.0 37.7 2.0 : 2.1 : 744.6 71.6 35.5 35.3 2.0 2.0 -- ran : 67.6 29.6 2.3 eeetae Venezuela . 65.7 42.7 . 1.5 ; Puerto Rico ee ees 65.6 21.7 3.0 : 65.2 20.9 2.2 ' 58.0 25.3 2.3 " 54.6 37.5 1.5 i 54.3 26.3 2.1 : 52.5 24.1 2.2 ; 40.9 20.2 .2.5 . 13.9 24.1 2.0 i 48.8 29.0 1.6 } 47.3 24.8 1.9 : 46.8 20.8 1.6 i 46.6 20.0 2.2 ; 44.9 22.0 2.0 44. 19.6 2.3 3 43.2 20.8 2.1 | 41.9 38.4 38.0 35.2 17.0 20.8 18.3 21.3 2.5 I 1.9 f 2.1 ; 1.7 32.9 15.0 - 2.2 i 32.0 12.S 2.5 ' Australia United States States white cede ce eeer eeee e eee 31.2 23.0 15.7 20.5 10.8 8.7 1.5 i 2.7 i 2.1 from 3 JACKSON 025454 14 DISTRIBUTION OF GASTROINTESTINAL CANCER CANCER 11881 1 Rate Rate Rate per 100,000 100,000 100,000 Male rate from 3 128. 5 148 110 ay Sl rit 8 | St. | 11 10 E to . 2 2 40 17? x 3 9 3] os 0 2. 2. 19 2. 2. 2. . 2. 2. 1. 2. 1 1 1 1. G 23.4 1.5 1.5 | | | : os : : : JACKSON 025455 : ' e : a TABLE 7 Figuur CANCER OF STOMACH MALEInternational List No. 151 asterisks divide columns into low intermediate and bigh ranges - Shown on the Map ( and II 29.4 29.4 England and Wales 29. from standardized 10,00 USA 1. stAgeandardized mortality males rates per Ivelan Northeast Nonwhite 27.4 27.0 Ismael 4Russian Russian Kazakh Soviet SSR FSK 12.0 112.0 91.6 USA NorthNorth Central Nonwhite Jagran SSK 79.7 elabor 79.3 Nova Scotia Candi . Bogota Colondaia Byelorussian SSR 78. 3 77.7 USA West Nonwhite New Zealand fechand fechand Purkmen Purkmen SSR Estonian SNK 75.3 75.3 USA South Nonwhite 71.9 Alberta Canada 9.3 Australia Kineitau Preun Brazil 67.3 New 09.2 Brunswick Canada 67.3 British British Columbia Canada GG Chaunnuda Azerbaydzlum SSR City Guatemala SSR N^thuanian SSR Linn Peru Noastrian Finland Romania Caracas Venezuel~ Phila Brazil USA NorthCentral Central White USA West White USA South White Frequncy ratios Frequency ul ratios Percentages Percentages of cancer in malmales es Christa Rica Antigua Antigua Yugoslavia Ilong Ilong Kong Ghana South Vietikon Surinaa | lawadi Pili ino Pilino Hawi Hawaii Caucasian Caucasian PapiaPapia and New Guinea Thailand Thailand Tran Morocco Philippines U.A.RPhi.lippines Kyadundo Kyadundo Uganda Bulawayo Khodesia ex W. Africa states Indonesia Lebanon Tunisia Cluba Fr Equ Africa states Kenya Madagascar Clungo Minshasa Mozambique Mozambique Gleylon 13. Iowa USA USA rural rural 12.5 11.4 11.2 11. 10.7 10 } 6,5 4 Singapore E. Indian 6. Muzambique 5.4 Kyadando Chigunda Chigunda 5.4 Singapoer Malay 4. U 4. 4.1 IV Supplementary frequency ratios 1. 4. NIGHT 4. 4.0 California USA Japanese 31.7 4. 3.9 PROBABLY IHCH 3.7 3. Japanese 3.3 3.0 Cluyana 1 Oriya Peru 1.2 Western Samoa 0.1 INTERMEDIATE Not shown on the Map III and IV Hungary Hungary 9999966 OOOOOLOGSPOT O O OLOGSPOT OOOOOOOOOOLOOGSPOT OOOOOOOOOOLOOGSPOT Congo Kinshasa 22.9 [ standardized incidence rates area Japanese 100,000 males 39.8 39.8 Japanese 22.5 per Chinese Rico Plata Argentina Newfoundland Canada 39. 36.6 36.6 , Hawaiian Arabia Cura^ao Fiji E. Indian Shanghai China Korea Colombia 20.1 Yugoslavia 19.8 Carchoslovakia 19.5 cities Nonwhite 18. Denmark 109.3 109.3 California 05.0 5846 5846 California 51.0 45 USA Malaysia Malaysia Malaysia E. Indian California USA Chinese USA Indian 39. G 36.2 LOW 17.2 Norway 35.5 * 16.6 USA 10 cities White Malaysia Malay 4.7 i North Vietnam Jamaica 16. 16.0 Kacbaw Ha Hoe Puerto Puerto Rico PROBABLY LOW 15 Turkey 13.7 13.5 New ZealandZealandZealand Algeria Assam Tudia JACKSON 025456 FIGURE 1 STOMACH MALE STOMACHSTOMACH MALE MALE#5) |, OCaucasian O filipino 'Japanese and Hawaiian from 4 oO: oO at or i os , ee a . / es : re) ~O O-.5 Seg 3. TO ; Oo oO o a - ae 1 MORTALITY RATE FREQUENCY RATIO A ee - oo J LOW INTERMEDIATE HIGH LESS THAN 23 0 TO o 40 AND OVER| : it LESS THAN 16 16TO 24 25 AND OVER [0] Qo W USA WHITE USA NONWHITE 5 ft 7 . Indion - 16 JACKSON 025457 17 Among white males in the United States between 1950-67 5 Table 8 the lowest rates were found in South and North Carolina 9.76 and 9.90 and the highest rate occurred in North Dakota 20.77 As is seen in Table 4-6 the sex ratio male rate per unit female rate is approximately 2.0 i.e. the rate of gastric cancer is twice as high among males compared with females in most countries When the sex ratio however is correlated with the specific site of the cancer within the stomach or with age considerable variability becomes evident Sterup and Mosbech 6 studied 1,983 consecutive cases of stomach cancer treated in Copenhagen hospitals during the period 1940-57 of which 1,212 diagnoses 61 per cent were established histologically and 771 39 per cent were determined by other means ray during surgery or at autopsy Of these 1,983 cases 43 per cent were in the pylorus 36 per cent occurred in the corpus 12 per cent were found invading the entire stomach 5 per cent were confined to the cardia and 4 per cent were in the fundus They found that the sex ratios varied considerably by site While the overall ratio was 1.7 the specific ratios were as follows cardia - 4.0 fundus 3.8 corpus - 1.8 pylorus - 1.5 and entire stomach - 1.2 MacDonald 7 studied 51 cases of adenocarcinoma of the cardia and compared their clinical features with those of 169 patients with cancer elsewhere in the stomach He found the following sex ratios cardia - 7 other 2.3 Flamant et al 8 reviewed 298 surgical cases of gastric cancer and found a sex ratio of 3 for the fundus cardia and corpus while the ratio for the pylorus was 1.4 Griffith 9 found that the sex ratio based on mortality rates varies by age He studied mortality rates in 24 countries for the six- JACKSON 025458 TABLE 8 MALIGNANT NEOPLASM OF STOMACH ICC 151 RATE 2007 11.24 1 50 1764 17892 . 2005 4049 10.14 10.14 14.09 14.09 16.72 12.76 13.41 11.49 11.87 13.35 10.94 13.13 17.66 17.87 10.44 11.14 12.45 15.36 15.03 14.48 16.60 + 10.92 17.56 NONZHITE MALE WHITE PEMALE FEMALE NUMBER RATE NUMBER RITE RATE 1422 19.08 1324 5.39 047 9.92 156 19.48 625 6.26 65 9.52 727 19.67 1044 6.19 366 9.39 1974 24.00 11296 7.17 787 9.92 90 22.66 1229 5.95 43 10.97 159 26.59 2471 8.70 50 7.11 141 31.12 270 7.08 50 11.34 605 1303 1570 27.72 22.35 21.12 377 2605 1379 6.70 4.97 5.05 C25 540 375 10.79 9.61 9.28 19 21.38 422 16 22.51 1757 26.39 9045 2.42 605 11.84 463 24.17 3146 5.27 153 9.95 50 23.27 2510 6.80 25 9.99 180 20.81 1430 5.5 72 7.78 480 22.39 2390 31.92 3 10.20 945 27.40 2132 6.91 1334 5.50 1016 0.41 1825 7.18 231 1151 3 432 9.76 13.53 9.70 11.81 - 217 1099 55 1595 817 33 66 24.42 24.39 18.26 22.24 24.77 21.38 25.07 6312 9.21 5743 8.34 3517 9.23 876 3581 514 1404 107 10.Ga 495 10.52 24 3.09 908 12.27 403 11.04 13 10.03 26 9.79 26 78.05 5 46.18 816 25.15 132 . < ' 6281 13 11.78 2 20.04 . 437 11.49 . . 14.05 559 25 8.54 8.54 td 5 18 18 JACKSON 025459 NEW YORK NORTH CAROLINA NORTH DAKOTA . ONIO UNITED STATES from 5 28441 2487 1306 12259 2027 2583 17603 1555 1007 1170 2807 7955 017 556 2568 4199 2399 7151 306 222524 10.52 9.90 20.77 15.10 12.70 34.12 17.15 19.30 9.76 16.48 10.91 12.01 13.43 13.67 11.04 15.35 14.31 18.15 13.82 15.22 TABLE 8- cont'd 2475 27.43 1406 20.20 12 18.57 1354 25.32 372 18.35 50 19.61 1804 29.39 . 33 23.55 9136 35 979 2002 23.45 . 19.87 20.49 21.45 46 36.51 1 1465 191 239 02 8 2.00 24.60 28.22 20.54 20.54 18.77 15.88 33606 24.03 . 18450 1655 719 7387 1669 1342 10795 1053 601 617 1846 5124 499 345 1699 2209 1345 4300 177 137254 9.78 5.32 11.70 7.42 6.32 6.62 8.62 9.97 5.09 0.30 5.89 6.42 7.03 6.53 5.74 7.21 7.27 9.51 4.59 7.70 ' 1400 729 7 500 247 25 733 15 584 25 515 1023 35 539 95 05 45 5 16725 12.59 9.07 9.91 10.62 11.00 10.34 10.66 9.51 1 .52 15.53 9.51 10.01 15.63 9.20 9.95 . 19 19 JACKSON 025460 20 year period 1958-63 Overaltl he mean ratio for all countries was approximately 1.0 for the youngest age group 25-29 rose steadily to a maximum of 2.23 at age 55-59 and then declined steadily to 1.46 at age 80-84 While these data are sectional Griffith cites evidence that the same trend can be seen in cohort studies as well and is therefore not a cohort phenomenon This sex ratio pattern was remarkably similar in all 25 countries analyzed and was not seen in cancers in other parts of the gastrointestinal tract Griffith attributes this sex ratio variability to calorie consumption and states that the available data suggests that the sex ratio of total calorie consumption would fit the observed pattern after allowing for an appropriate latent interval JACKSON 025461 SOCIOECONOMIC OCCUPATIONAL RELIGIOUS AND ASSOCIATED FACTORS Numerous factors some of which are of etiologic importance influence gastric cancer incidence rates Haenszel 10 conducted an exhaustive review of available data up to the year 1953 and looked at variations in standardized stomach cancer incidence and mortality in relation to demographic characteristics socioeconomic status urban residence geographic latitude and nativity within the United States He found a race ratio nonwhite rate per unit white rate of 1.33 for males and 1.28 for females with this difference persisting in all regions of the country as well as in metropolitan and metropolitan areas There were very small rural differences with the urban rates generally slightly higher The greatest urban differences were found in areas with central cities Los Angeles New York Chicago and Cleveland Haenszel points out however that this may be due to the fact that these cities have sizeable foreign groups This small urban disparity has been duplicated in England and Wales Norway and Finland Levin and associates 11 also found very small urban differences for both males and females in New York and in Connecticut and Iowa as well Stomach cancer mortality rates Haenszel found were higher in Northeast and Northcentral states compared with rates in the South , and the rates appeared to vary with latitude This appears consistent with the observation that countries furthest north of the equator have JACKSON 025462 the highest rates as is seen in Figure 1. Within the U. S. this pattern also seems to hold based on Hoover's et al 12 mapping of gastric cancer mortality by county Based on three studies Haenszel also concluded that the lower social classes exhibited the highest rates Indeed one study found this pattern in cases among married women classified by their husbands occupation Gregor et al 13 reported on a country ecological study where the investigators examined the mortality rates for all cancers against the rates for stomach and intestinal cancers and correlated these rates with the standard as measured by the intake of animal protein They found that while cancer in general was not correlated with protein intake there was a strong negative correlation between protein intake and stomach cancer for both sexes This again indicates that the higher social classes have the lowest rates The authors however express the possibility that living standards may be related to other factors which influence the etiology or pathogenesis of cancer Newell and Waggoner 14 determined the correlation between adjusted stomach cancer mortality rates and two indicators of socioeconomic status - per capita income and number of physicians per 100,000 population They found negative coefficients of 0.87 and 0.77 for white males and females respectively for the former and coefficients of 0.182 0.77 for the latter They also correlated the death rates with the temperature index mean annual temperature plus range and found a very high negative correlation a relationship | which they suggest is not of etiologic importance and which may indicate a secondary association with social class Davies and Griffith 15 in their study of the relationship between soil and cancer to be JACKSON 025463 23 discussed later also found that gastric cancer mortality rates were inversely correlated with social class Numerous investigators have examined in depth the association between social class and stomach cancer incidence Cohart 16 studied all cases of carcinoma of the stomach in New Haven Connecticut between the years 1935-49 Each case was assigned to one of twenty- four ecological socioeconomic districts based on residence The districts were delineated according to a determination of the character | of the dwellings in the area along with other factors such as family income occupation dependency delinquency religion race and nativity A total of 658 cases was collected during this period - 380 among males and 278 among females although 22 were not available for study The 24 districts were condensed into seven fairly homogeneous socioeconomic areas which ranged from rich to poor Cohart found that the observed male rates were significantly lower than expected in the richest area and higher than expected in one of the middle and in the lower class area When these seven areas were further reduced to three well middle class and poor the difference between observed and expected number of cases among females residing in the poor area became statistically significant but significant differences for males disappeared a finding which is in disagreement with the premise that social class and stomach cancer incidence are inversely related . ; . Graham et al 17 examined the records of all patients with cancer who were residents of Buffalo New York and whose cases were reported to the Erie County Cancer Registry during the five year period 1948-52 Socioeconomic status was determined on the basis JACKSON 025464 24 of the median rental in the patients census tract of residence according to the U. S. 1950 census The 72 census tracts of Buffalo were divided into quartiles with quartile I having the highest median rental The investigators found that there was an increasing age- standardized stomach cancer incidence rate with decreasing socio- economic status for both sexes Among males the age standardized incidence rates per 100,000 increased from 19.6-24.9-28.0-38.19.6-240.9-28.0-38.0 among those in quartile I to IV Among females the rates increased from 11.8-14.5-13.7-23.0 They also attempted to explain Cohart's inconsistent findings with the observation that New Haven had an unusually large proportion of Americans most of whom were in the lower social classeansd that this ethnic group had extremely low rates of gastric cancer although Italy is reported to have rather high stomach cancer mortality rates Seidman 18 examined the adjusted mortality rates among the Puerto Rican white population of New York City between 1949-51 for those aged 25-64 The rates were classified jointly by sex religion and socioeconomic status The latter was defined according to the socioeconomic status of the Health Area of residence of the deceased These Health Areas are combinations of contiguous census tracts of about 22,000 persons Table 9 presents the results of the study As is seen there was among males an increasing mortality rate as social class decreased but this pattern not hold among females When social class was classified by religion there were increasing rates among males within each religious group as social class decreased from high to low Among females however this trend was inconsistent Interestingly there was not much difference JACKSON 025465 25 . : . ; - . : va we es penne ce FEMALE 1 COMPAMED WITH j \ r r THE CANCER OF THE re 1373172.7 3 7 1326 332.5 1948 1440 | 9 | 2 1 ) ; Ci -t a ia { Tv 6 , i 2 L i. . : r 732 , 5 q 1+ i 443 443 F 3. 150 eae | cod . 6 tL" waites by Scorcon Clara - HOM MICGLE LOW MIG- a HIGH NJ MICH c MIGA white by 0048 Cand Cand F6 347 MIDONE MIDDLE STCOIN HIDDLE J NJ , P LO LON NJ 4.0 UA ty While Groub : =a inal Ss can ri <A. ; H : t 264 269.9 921 921 : 251.2 240 251.2 1.8 : 27532499 27532499 14.2 i - : 3 $ 150 . 152 1783 10.9 . 933109 11.0 592 10.8 .- , 71 716 12.5 159 1806 11.5 . 61 121.6 121.6 72- 72- 54 58.0 102 : 1140 . 21 1653 1 134165161 49 134165161134165161 0.0 134165161 Z.E 277249.6 107 342 151 277249.6 5344 347.2 187.2187.2 12.7 16 235 584 1.6 : 132 2 10 Frc 3. Cancer of the esophages and of the stomach from 18 JACKSON 025466 26 between the rates of males and females of different religions although one might expect the rates among Jews and Catholics among whom there are a greater proportion of first and second generation immigrants to be higher than the rate among Protestants In fact male Jews had the lowest rate of any religious group while females had the highest although the range was rather small Newill 19 also studied cancer mortality mortality among white residents of New York City He investigated a total of 83,341 deaths of which approximately 10 per cent 8,566 were ascribed to stomach cancer Examination of religious afilliation was restricted to those aged 45 and over and rates were not age adjusted on the assumption that variations in the religious distribution of the population by age is inconsequential after age 45 an assumption which must be seriously questioned Newill found In addition tests of significance were not presented that for each sex separately and for both sexes combined Jews had the highest average annual death rate from stomach cancer followed by Protestants and Catholics although the differences were not very great These conclusions therefore are in partial agreement | with those reported by Seidman Buell Dunn and Breslow 20 analyzed 10,401 male cancer deaths in California among those aged 20-64 during the period 1949-51 . Each death was assigned to one of five social class levels based on the reported occupation on the death certificate The occupational social class levels based on intuitive judgements " were developed by the National Office of Vital Statistics The five categories from highest to lowest are professionals intermediate sales clerical and skilled semiskilled and unskilled Farmers were excluded from the JACKSON 025467 27 classification The investigators found a strongly increasing stomach cancer mortality rate from Social Class I to V indicating a definite inverse relationship They also found nine occupations with an excess number of deaths * based on Standardized Mortality Ratios significant at the 0.05 per cent level Social Class III - bakers masons plumbers toolmakers Class IV - meatcutters and cooks Class V - lumbermen laborers longshoremen Significant deficits were found in Class I - accountants Class II - teachers managers officials proprietors in trade in manufacturing and in all other industries Class III - other clerical workers Kraus Levin and Gerhardt 21 in an excellent hospital control study examinethde relationship between occupational exposures and gastric cancer among males aged 25-74 admitted to Roswell Park Memorial Institute in Buffalo N. Y. between November 1948 and September 1951. Occupational histories were obtained by personal interview and included all occupations in which the patient was ever employed In addition information about exposure to a specific list of physical and chemical agents were elicited from the patients This was followed by the compilation by a team of industrial chemists of a probable list of exposures to chemical agents based on the job descriptions reported Information on a total of 56 stomach cancer cases was reported Four control groups were selected 1. patients with digestive cancers other than stomach or esophagus 2. digestive cancer patients 3. tumor diagnoses of the digestive system other than esophaguosr stomach 4. other tumor diagnoses A total of 667 controls was JACKSON 025468 28 selected in this manner Each control group was matched with the case on age and population size of residence apparently by group frequency matching to eliminate the possibility that age or urbanicity would be responsible for any differences found There were 18 broad occupational groups Cases and controls were compared regarding the proportion who had ever worked in any of these groups and regarding the proportion who had ever been exposed to the specific physical and chemical agents Duration and concentration of exposure were also examined Only one occupational group was significantly associated with stomach cancer - the products occupation - in which there is exposure to iron dust The investigators determined that 66 per cent of the cases and 43 per cent of the controls had ever worked in this industry a difference which was highly statistically significant 0.002 It should be noted that the proportion of both cases and controls ever employed in this industry seems exceptionally high and perhaps the possibility of selective admission to Roswell Park should have been considered At any rate Kraus et al found that 12.5 per cent of the cases had been exposed to iron dust for 10 or more years against only 1 per cent of the controls Subsequentto year employment involving exposure to metals 12.5 per cent of the cases had been exposed to grain dust for 5 or more years usually as a farmer compared with 1.5 per cent of the controls although the authors suggest that the relationship between stomach cancer and farming life is probably dependent on some factor other than grain dust itself Additionally of all stomach cancer cases 17.9 per cent were exposed to inorganic dust with free silica for 10 or more years against 7.8 per cent of the controls These individuals were not included in the JACKSON 025469 iron dust and grain dust exposure groups Nosignificant differences between cases and controls were found when analyzed for smoking habits and marital status but almost four times as many cases 23.2 per cent vs. 6.4 per cent had been born in Poland p 0.0001 a finding which was independent of occupational factors Sigurjonnson 22 examinethde death certificates of all adult males between the ages 35-64 who died in Iceland during the period 1951-60 A total of 486 deaths due to gastric cancer were identified and proportionate mortality rates which apparently were not standardizweerde calculatfeodr five socioeconomic classes into which all cases were categorized I -farmers II- labourers III- seamIeVn= craftsmen :- white collar workers The proportion of all deathds ue to stomach cancer decreased progressively 25 per cent - 19 per cent - 15.5 per cent - 9.9 per cent - 8.4 per cent although it is difficult it would seem to infer that seamen for example belonged toa higher social class than farmers or laborers The author attempts to correlate all differences in these classes to differences in dietary consumption of 3,4 benzopyrene an attempt which may be of dubious value when based on proportionate mortality rates which additionally have not been standardized JACKSON 025470 MIGRANT AND ETHNIC STUDIES Lombard and Doering 23 in 1929 made the first attempt to analyze and compare cancer mortality rates for foreign and native They studied cancer deaths in Massachusetts during the year period 1920-24 adjusted to the 1920 state population Rates at numerous sites including stomach were calculated separately for male and female foreign native of foreign parents and native of native parents Results for stomach cancer are presented in Table 10 and as is seen there is a definite decreasing trend fromforeign to native- born of native parents in each sex Surprisingly females had higher rates in two of the three nativity categories a finding which has not been duplicated and which must lead to a serious doubt about the accuracy of the data The authors tried to relate the differences found in Massachusetts on a nativity basis to the differences found in England on an economic social class basis Apparently no attempt was made to standardize for economic status which seems feasible since standardization was employed The attempt to explain the differences in this manner was supported by results of a survey which indicated that a greater proportion of foreign were poor when compared with the other two nativity groups native of native and of foreign parents When cancer death rates were analyzed by specific country of birth it was found that Russians and Scandinavians had the highest rates among males while Teutons Austria Germany Holland Belgium and Canadians JACKSON 025471 31 TABLE 10 . Stomach Cancer Mortality Rates per 100,000 in - . | ; Massachusetts 1920-24 by Nativity Nativity _ Male Rates Female Foreign born 35.5 30.5 Native born of foreign parents 20.8 22.4 Native born of native parents 11.8 14.3 from 23] . ; JACKSON 025472 32 had the highest rates among females a finding consistent with subsequent surveys The authors conclude that foreighan ve much higher stomach cancer mortality rates than native born of native parentansd that stomach cancer rates are abnormally high among both males and females in all foreign groups While there apparently were some shortcomings in collection and analysis of the data Lombard and Doering should be credited with first realizing the importance of ethnicity in cancer incidence Choi 24 investigated intestinal cancer mortality the general ethnic distribution of among the foreign in Manitoba gastro- during the years 1955-65 and presented SMRS based on the rates among native- born and all Manitobans from the 1961 census A total of 1,198 gastric cancer period deaths among foreign - 839 among males and 359 Manitobans was among females recorded during For males the this average - annual adjusted gastric cancer death rates among the foreign 100,000 - were almost twice as high as among natives - 17.7 100,000 Immigrants from among males for over twenty countries every country except were included in this study and the U. S. the SMRs were above born from Scandinavia SMR - 248 100 with the highest among foreign the USSR SMR = 221 and Poland SMR = 208 Graham and associates 25 studied all admissions with regard to ethnicity to Roswell Park Memorial Hospital between 1 January 1945 and 31 December 1956. Of the 35,701 admissions during this period 18,677 were diagnosed as having some form of malignancy A control group was chosen from admissions during the same period not associated with cancer and all findings are based on adjustment for variations in age occupation rural residence and tobacco use JACKSON 025473 .33 The investigators found that there were 21.4 per cent more foreign- born patients among the stomach cancer groups compared with the controls 0.0001 The relative risk of gastric cancer among foreign compared with native was 2.5 When Poles were analyzed separately they found that 22.4 per cent of the gastric cancer patients were of Polish origin compared with 9.8 per cent of the controls 0.0001 and Poles overall had a relative risk of 1.2 when compared against all other foreign Adjustments were made for the four factors mentioned above because it was found that the various ethnic groups differed substantially in the distribution of these factors which are generally closely associated with cancer etiology Haenszel 26 examined all cancer deaths which occurred among foreign and native in 35 states in the U. S. during this single year 1950 and computed site specific mortality rates for native white total foreign white and 12 major foreign groups England Iceland Norway Sweden Germany Australia Czechoslovakia Poland USSR Italy Canada and Mexico Age adjustment was by the indirect method and SMRS are presented based on three separate referents - U. S. native whites total whites and referents derived from foreign countries SMRS for gastric cancer based on U. S. native whites were statistically significantly elevated p 0.01 for total foreign whites and for foreign whites from each of the 12 countries used in the study and the SMRs were elevated for both males and females In fact stomach cancer was the only site in which the SMRs were consistently elevated Among males they ranged from a low of 115 among Mexicans to a high of 171 among Czechs Among females the SMRS ranged from 130 among Canadians to 264 among JACKSON 025474 34 Czechs Overall the highest risks were found for migrants from Poland Czechoslavakia Norway and the USSR Many investigators both in this and other countries have examined cancer incidence and mortality Polish migrants especially in relation rates among Japanese and to stomach cancer for which the two native countries have rates among the highest in the world Smith 27 studied the mortality experience for the years 1949-52 of Japanese residing in Hawaii and the continental U. S. based on information provided by the National Office of Vital Statistics Both SMRs based on the indirect method of adjustment and rates derived from a direct adjustment were calculated but since they were generally similar only SMRs were presented based on two different referents - the white and white U. S. populations When both male and female cancer rates among the Japanese were compared against the white and white U. S. populations on which the SMRS were based stomach cancer SMRs were significantly elevated p = 0.01 Gastric cancer was the only one of approximately 15 sites studied for which the SMRS were significantly high in all four comparisons For both males and females separately the SMRs were approximately 300 when U. S. whites were used as a referent and 200 when compared with U. S. whites These calculations were based on 395 male stomach cancer deaths and 125 female deaths Buell and Dunn 28 studied the Issei first or immigrant generation Japanese and Nisei second generation in California Among the Issei there were 16,377 males and 20,271 females , while the Nisei included | 63,373 males and 59,424 females Cancer deaths during the period 1949-62 were analyzed in two separate year calendar periods centered JACKSON 025475 35 around the years 1952 and 1959 to enable use of census data for 1950 and 1960. Generally it was found that Issei males had gastric cancer death rates closer to male rates in Japan where they are exceptionally high than to the rate among white Californians Male Nisei on the other hand had rates which were closer to those found among Californians SMRs were calculated separately for males aged 45-64 and for those 65 and over with the expected number based on the rates among California white males Based on this referent the SMR among those aged 45-64 for Japan was strikingly high at 840 while it was 380 for Issei and 280 for Nisei In the 65 and over group the SMR for Japan was 530 and 480 for the Issei no data were available for the Nisei The authors suggest that differences in diet or other environmental factors may account for some of these findings and rule out the possibility of bias due either to under- or diagnosis or false information about birth place Staszewski and Haenszel 29 studied and specific cancer mortality rates among the 860,000 Polish in the U. S. using Haenszel's technique described earlier Rates among these migrants | were compared with rates among U. S. native whites and with native Poles The results indicated that gastric cancer mortality rates for both male and female Polish migrants were quite close to the rates in Poland among those aged 35-75 The authors state that the persistence of characteristic dietary customs among ... ethnic groups in the United States assumes significance in this context Staszewski et al 30 also studied the cancer mortality of Polish migrants to Australia during the period 1962-66 and compared JACKSON 025476 36 the specific mortality rates of this group for those aged 40 and over with the rates in Poland and Australia for the period 1964-65 Among males the migrant rates were intermediate between the high rates in Poland and the low rates in Australia for the entire age range 40 and over Female rates were intermediate beginning at about age 50 the but between 40-50 rates were the same as those in Poland Calcula- tions were based on 59 male and 26 female deaths and quarters of the deaths occurred among those residing in Australia for a period of 7-18 years The intermediary position found here differs from a similar study by Staszewski and Haenszel 29 cited above where migrant rates were not intermediate but were rather almost the same as Polish rates Staszewski suggests that since the former study pertained to 1950 rates while this study involved 1962-66 rates there is a possibility that migrants risks changed with time He also suggests as possible explanations for the different findings the following differences in length of residence in the host country differences in the environment of the host country and differences in migrants backgrounds due to selection Monk and Warshauser 31 capitalized on a unique situation to investigate the relative importance of early environmental and constitutional factors on the incidence of stomach cancer Generally when migrants are studied it is assumed that they have permanently left their country of birth and the finding that their mortality or incidence rates lie closer to the host than the adopted country can be explained on the basis of an early environmental factor the effects of which linger despite the transition There is in New York City a JACKSON 025477 37 large Puerto Rican population however which is unique because many of them return frequently to their native country and stay for unknown lengths of time As a group therefore they are being continually exposed to the environmental factors prevalent in Puerto Rico where mortality rates are higher than in New York City Deaths among Puerto Ricans in New York and in Puerto Rico and among the white and white populations in New York City were examined during three periods 1958-62 1963-67 and 1968-71 Apparently an indirect standardization was used based on 1960 and 1970 census figures and on an intercensal estimate and admission criteria were based on death certificate information The results indicated that the adjusted mortality rates among males decreased during the three periods under study in each of the four population groups while the same trend was seen among females except for Puerto Rican born women whose rates remained rather stable In all three periods also the rates among Puerto Rican migrants to New York were lower than among Puerto Rican natives and higher than among native whites in New York These results parallel to a degree the findings of other migrant investigations where the country of origin experiences rates higher than that of the host country In this study however the rates New among Puerto Rican migrants were closer to the rates in York than . to those in Puerto Rico indicating that the occurrence of this disease is more responsive to recent rather than early environmental exposures despite the fact that Puerto Ricans tend to return to their homeland often for extended periods Other possible explanations offerebdy the investigators for these unexpected findings include differences in diagnostic procedures and treatment differences in resistancteo JACKSON 025478 38 dietary changes selective migration of Puerto Ricans at lower risk of developing stomach cancer and selective return of migrants to Puerto Rico before retirement or death The best explanation according to the investigators however is the recency of migration and the young age of Puerto Rican migrants which seemtso account rather incompletely for the fact that rates among migrants were closer to those among native whites in New York rather than to rates in Puerto Rico Tulchinsky and Modan 32 studied the incidence of gastric cancer among Jews in Israel for specific ethnic groups between 1961-63 based on 1172 cases Israel is an ideal country for studying rates among different nationalities because of its exceptionally heterogeneous ethnic composition Cases were classified into those with definite and probable diagnoses The former were dependent upon either a mortem examination or on evidence from operative procedures with a biopsy while the latter were based on explorative surgery rays clinical signs and symptoms or death certificate information The overall mean annual incidence was 100,000 among males and 100,000 among females while the sex ratio was 4 for those aged 35 or younger and 1.6 among those over 35 years of age The cases were then classified into one of three population groups immigrants from Europe Africa and Asia Analysis of the specific rates revealed that male European immigrants had higher rates in every age group except among those 75 and over The overall incidence rates were indirectly adjusted to the 1961 Israel population and ratios of these standardized rates are presented although they are not really mutually comparable The adjusted rates among males indicated a distinct gradient Europeans = 25.5 Africans = 20.8 JACKSON 025479 39 Asians = 12.4 native Israelis= 8.7 When these were further analyzed according to specific country oforigin immigrants from the following countries had the highest rates Rumania Germany Austria Morocco and Bulgaria All but Morocco are eastern European countries and these results therefore agree with the findings of other investigators On the basis of these ethnic differences in gastric cancer incidenacned mortality and on other analyses not mentioned here Terris and Hall 33 considered the hypothesis that the decline in mortality from stomach cancer in the U. S. may be at least partly attributablteo the change in the population composition specifically the declining rates may be a result of a decreasing proportion of foreign born among whom the rates are highest within the U. S. Accordingly they examined the specific mortality rate trends among three population group - white native white foreign and Negroes M in residents of New York City for two triennial periods 1949-51 and 1959-61 The specific mortality rates among both sexes in each of the three population groups declined during the decade under study and the adjusted mortality rates revealed that males had considerably higher rates than females in each of the three groups with Negroes experiencing the highest rates followed by foreign and then native whites Based on the declining rates within each of the three population groups and on other pertinent information the authors conclude that ' the declining mortality rate within the U. S. can be attributed neither to a decrease in the proportion of foreign nor to improvement in therapy and probably not to improvement in diagnostic JACKSON 025480 40 accuracy The decline instead is considered to be real and probably be explained by a decreased exposure to some unknown can environmental or dietary factor 20-30 years the incubation period prior to mortality or alternatively by an increasing exposure to some protective factor In addition to these international migration studies there is at least one national study of migration from areas in the southern in region of the U. S. to the state of Ohio Mancuso . 34 this study reports that white migrants to the state from the South experienced stomach cancer rates which were approximately twice as high as expected 28 observed and 13.7 expected Unfortunately not enough discussion concerning the many selective factors involved in migration is presented which leads to a less than convincing argument JACKSON 025481 sat DIETARY FACTORS Dietary factors are perhaps the most attractive explanation for the vast differences in international gastric cancer incidence and mortality rates and for differences between groups with different attributes within countries After all all people have the potential for exposure to presumed carcinogens in food especially in view of numerous animal studies which have indicated that many food additives and preservatives are indeed carcinogenic either in their primary form or after metabolic or bacterial conversion Dietary factors are also tempting explanations for the results of migrant studies which generally indicate that first generation migrants experience rates closer to their home rather than host country while second and third generation offspring gradually approach the rates of the host country It is reasonable to assume that the latter also gradually alter their diet as they are assimilated into the host country culture The problems in assessing an accurate dietary history in retrospective studies however are overwhelming If the latency period is assumed to be on the order of at least two or three decades prior to which there must have been sustained exposure to the putative carcinogen the investigator must precisely estimate the specific dietary components quantitatively and chronologically Since this is rarely if ever possible qualitative estimates are often accepted instead While even the latter can generate leads and hypotheses it is JACKSON 025482 42 _- difficult to accept these results as strong evidence for implicating a specific food or carcinogen as being of causative importance in the etiology of gastric cancer despite statistical significance These problems are compounded by the fact that investigators generally do not provide in scientific papers the questionnaire used in their studies so that a comparison of results is often difficult and may in fact be meaningless Naturally it would be ideal to conduct prospective studies to determine the relationship between dietary consumption and subsequent development of stomach and intestinal cancers although opportunities for this type of study are at best infrequent Perhaps this type of investigation could be incorporated into large intervention trials such as the LIPID study being conducted by the Department of Biostatistics and Epidemiology at UNC if participants were relatively young at entry and if these individuals have a high compliance rate with the dietary demands for a long period of time subsequent to inclusion in the study Despite these shortcomings of dietary investigations findings cannot be ignored although they should be interpreted with great caution Bjelke 35 has exhaustively reviewed the literature on case- control dietary studies and its relationship to gastrointestinal cancer and he concluded that findings indicating any association between diet and these cancers ... were few equivocal and sometimes conflicting However he found the evidence of a negative association between gastric cancer and intake of vegetables fairly strong While a negative association between consumption of fruits and Vitamin C were found by two investigators they explained these findings as JACKSON 025483 43 being either a consequence of the gastric condition or as a secondary association with the primary relationship between socioeconomic status and gastric cancer Nevertheless no studies have indicated a higher intake of fruits or vegetables among stomach cancer patients compared with controls Intake of highly salted foods was the only dietary factor which was convincingly positively associated with stomach . cancer Bjelke also nicely summarizes the major methodologic problems | encountered in control dietary studies Most reports give little specific information about the actual dietary questions that were used and what time period they referred to few details are given on the selection of cases no mention is made of how they might correspond to all new tion and information the source population cases occurring in a defined populawhich might aid an assessment of how for the controls might compare to the general population on pertinent variables is almost equally scarce It should be reiterated however that methodologic problems and inconsistencies notwithstanding dietary hypotheses are still attractive and may explain variations in gastric cancer incidence in either a primary or secondary manner As an example of the former Shamberger et al 36 suggest that the declining gastric cancer incidence in the U. S. may be partly attributable to dietary factors the widespread consumption beginning in 1920 of wheat cereals rich in antioxidants and the increasing use of antioxidants as food preservatives which began in the last 1940's Antioxidants act by terminating the propagation of chain reactions which accelerate the oxidation process and Shamberger suggests that antioxidants applied near the site of carcinogenesis may prevent peroxidation and attachment of the carcinogen to DNA which may be one of the rate limiting steps in carcinogenesis JACKSON 025484 44 Some of the most common antioxidants are BHT butylated hydroxytoluene and BHA butylated hydroxyanisole as well as lecithin selenium and tocopherols While gastric cancer rates have been declining steadily since 1930 the author points to the deflection in the curve in 1947 hearlding an even more rapid decline This he implies could be explained by the introduction of the antioxidants BHA and BHT as food preservatives In the spirit of an ecological comparison , Shamberger examined 14 male cancer sites in 17 paired cities with the latter dichotomized into high and low selenium consumption areas but the method of assignment of cities into high or low selenium areas is not discussed He found that for all gastrointestinal cancerthse high selenium areas had . lower SMRs although the choice of a standard was not identified Correa and associates 37 on the other hand present a model for gastric cancer pathogenesis in which dietary composition plays a secondary role They postulate that intestinal type gastric carcinoma results from mutations and cellular transformations which begin during the first decade of life but which are mediated by abrasive or irritant food substances such as hard grains or salty foods which permit the carcinogen to pass the mucous barrier of the stomach if the synthesis of these substances is not initially by the presence of antioxidants JACKSON 025485 OTHER RISK FACTORS ABO Blood Groups Stocks 38 in a study of geographic variation in the incidence of stomach supply was cancer found that towns of moderate hardness had in England and Wales whose water lower standardized mortality rates than towns whose water was soft or very hard The former were mostly in southern and the latter in northern counties Aird et al 39 however felt that the true correlation may not be with water hardness but rather with genetic differences between northern and southern provinces as reflected in the frequency of ABO blood groups They drew cases of stomach cancer from a number of hospitals in England and Scotland whose diagnoses in most instances were based on histological evidence but in a few cases the diagnosis was based on histologic findings An unspecified number of controls was obtained either from a section of the patients from the hospital concerned or from local transfusion centers which putatively represent the general population of the locale from which the patients were drawn While they did not prove their original hypothesis they 0 found that blood group A was more frequent and less frequent in OS patients suffering from stomach cancer compared with the blood group S E distribution based on the control series found in the general population of the same locality Buckwalter et al 40 conducteda control study with cases a drawn from six hospitals Individuals were included in the study JACKSON 025486 46 if both their blood type and a histological diagnosis of stomach cancer were available Overall they found an excess of type A among cases - 52.1 per cent which was significantly different from the proportion of type A among controls 47.6 per cent who were selected from among healthy blood donors at either of two hospital blood banks There was no apparent matching for sex and age which admittedly may not be confounders with respect to blood type nor was there any matching for ethnicity which may be a confounder There does however seem to have been an attempt to control for these factors in the analysis | although sufficient data are not presented Some investigators have broadened blood type analyses to include the site of the cancer within the stomach | Billington 4 attempted to demonstrate that carcinomoaf the stomach is associated with differing ABO blood groups dependent upon the site of the cancer in relation to the three major types of gastric glands He studied 483 cases of adenocarcinoma of the stomach excluding those which occurred among immigrants at the teaching hospitals afilliated with the University of Sydney during the period 1947-54 All diagnoses were confirmed microscopically and classified according to involvement of the three major types of gastric glands based on the site of the lesion - cardiac body and prepyloric He then like Aird 39 compared his results with the ABO blood type distribution of blood donors from | . the same geographic area His analysis indicated that there was no overall association between blood group A and carcinoma When further analyzebdy site | however there was a significant association between type A and pr^'pyloric and cardiac lesions while the body and fundus were JACKSON 025487 47 significantly associated with blood group 0. Billington attempts to explain Aird's findings of an association with type A by suggesting that Aird may have had a greater proportion of prepyloric and cardiac lesions in England along a generally greater percentage of blood type A's compared with Sydney which from other evidence may be true White and Eisenberg 42 reported similar findings They examined 707 caseosf gastric cancer and categorized them by blood type site sex age and place of residence Sites were determined from hospital records pathology and radiography reports or from findings at surgery They found that blood group A was commoner than in the series as a whole with the excess of A most marked among males with cancer in the pyloric region These site specific analyses may help explain negative findings by other investigators such as those of Crumb and Willets 43 who concluded on the basis of 123 cases of gastric cancer that there was no association with blood group A. Tobacco and Alcohol Flamant 44 and his group in France has examined the relation between smoking and alcohol use and the incidence of gastric cancer One survey included 64,834 cancer cases recorded between 1943 and 1960 and the investigators found that in general the higher the correlation between a site and the use of alcohol and tobacco the higher was the standardized sex ratio In addition to this large survey the study included a series of 298 surgical cases of gastric cancer selected with great diagnostic accuracy although the selection criteria are not offered They found that the sex ratio for the fundus cardia and corpus was 3 while it was 1.4 for the JACKSON 025488 48 pylorus The authors imply that since females smoke and believed to drink more infrequently than males a strong association between smoking or drinking and cancer will result in a higher sex ratio This 1 sex ratio would therefore provide evidence for an association However, in another investigation conducted by this group 45 the conclusions differ The study was a retrospective analysis of 3,937 male cancer patients matched for age and interviewer with two controls - one healthy and the other ill but not cancerous The authors found no differences between stomach cancer cases and controls in the proportion of smokers total amount smoked and in the proportion of cigarette smokers who inhaled the smoke Hammond 46 however did find some differences based on a prospective study of over one million persons of whom 422,000 males between 40-84 years of age were followed for approximately three years This group included smokers and smokers and the investigators determined that the stomach cancer mortality ratio for smokers non- smokers among males aged 40-69 and 70-89 was 1.39 and 1.38 respectively For gastric ulcer however the corresponding mortality ratios were exceptionally high - 5.38 and 2.97 respectively It would seem that the smoking ratios of peptic ulcer and gastric cancer cases should be similar ulcer is a precursor unless ulcer patients who smoke selectively experience a higher mortality rate than do smoking ulcer | cases Wynder 47 as chairman of a gastrointestinal cancer conference . in 1965 concluded that there appeatorbse no association between tobacco and gastric cancer Zacho et al 48 49 in a series of papers have attemptetdo refute this conclusion One of their studies JACKSON 025489 49 did not include a control group of any kind and while we are informed that the cases were drawn from admissions to the Finsen Institute in Copenhagen between 1948-69 no mention is made of the selection criteria or of the proportion of cases interviewed In another study apparently drawn from the same hospital there is an attempt to compare the proportion of cases who ever smoked with a survey of smoking habits conducted by the Danish National Health Service in 1952-54 The study was poorly designed and the comparisons are of dubious value since there was no standardization of information collection Nevertheless the investigators attempt to explain the sex ratio for gastric cancer which they determined from their study by the difference in tobacco consumption in the two sexes and conclude that this would seem to bea weighty indication of the aetiological role of tobacco Stocks 50 using an ecological design which he seems to prefer studied the adjusted mortality rates at different sites in 20 countries for the period 1905-65 and correlated these rates with average national consumptioonf four commodities cigarettes tea coffee and solid fuel He found no statistically significant associations between stomach cancer rates and any of these commodities although significant associations were found in both sexes between solid fuel and tea consumption and rates of intestinal excluding rectum cancer Stocks attempts to explain these findings by stating that onlya limited pro- portion of people are susceptible to gastrointestinal cancers but the most likely target in the presence of certain carcinogens seems to be the intestine There appears to be a conviction among some investigators therefore that there must be an association between smoking and stomach cancer notwithstanding the equivocal results JACKSON 025490 50 For alcohol consumption also there is no convincing evidence of any association This was the conclusion reached by Bjelke 35 in his thorough review and also by the Flamant 44 group cited above Graham et al 51 reached similar conclusions from a retrospective dietary study of 168 male and 68 female gastric cancer patients at Roswell Memorial Institute They matched cases on age country of birth and nationality with neoplastic gastrointestinal disease controls and found no relationship between gastric cancer and frequency of use of various alcoholic beverages The possibility should be considered however that alcohol may be of etiologic importance in both cancerous and cancerous gastrointestinal diseases which case one would not expect any differences between the two groups Soil and Fertilizer Davies and Wynne Griffith 52 studied cancer mortalitiyn Anglesey County England during the years 1943-52 in relation to soil type near the home of the decedent social class and water supply A total of 178 male cases of gastric cancer and 140 female cases were examined and expected numbers were calculated on the assumption that the proportion of cancers at a specific site having a specific attribute e.g. soil type was the same as the proportion of all cancers with that attribute Soils were classified according to two criteria nature of the parent material and seasonal productivitays determined by moisture content and drainage Water supply was categorized into two groups - public mains and other sources including wells springs and stored rainwater 7 . For stomach cancer highly statistically significant associations were found for soil social class and water supply There was a BATO JACKSON 025491 51 relative excess of deaths seen on soils which show great biologic and activity over a long season among those of low social class among those whose water supply came from other sources The general association between gastric cancer and soil remained intact even after adjustment for age sex water supply and social class although the association maintained statistical significance only among those 65 | years of age and older Tromp and Diehl 53 studied the relationship between major regional soil types and mortality from stomach cancer in the Netherl. a ben twdeesn 1946-52 A total of 14,072 deaths were studied in this manner and they found that in almost every age group those living near riverclay and ordinary sandy soils had averaged stomach cancer mortality rates below the national average while those living near clay peat and reclaimed peat soils had higher rates in every age group While the authors were unable to explain the observed relationships they expressed the opinion that it was at least to some degree of etiologic importance perhaps because of differences in trace element distributions within soils Unfortunately they make no mention of how their soil classification compared with that of Davies Stocks and Davies 54 examined the zinc and copper content of the soil in gardens of people who had just died of cancer cases and other diseases controls To qualify the person must have lived in that house for at leas1t0 years on the assumption of a latency period at least that long They computed the logarithm of the copper ratio and found that the average log of the ratio was higher in the gardens of people who had just died from gastric cancer after 10 or more years of residence compared against houses whose occupants had died of other JACKSON 025492 52 causes Similar analyses of other cancer types showed no differences between cases and controls While the investigators did not postulate any possible carcinogenic mechanisms involving copper or zinc they suggest that the copper ratio may be of importance and demonstrate that while there is no statistical association between zinc and copper soil content separately with potatoes grown on it theries a fairly strong correlation between the ratio in the soil and potatoes grown on that soil In a recent ecological study by Armijo and Coulson 55 there was an attempt correlate the use of nitrogen fertilizers in Chile with the stomach cancer mortality rates in different provinceosf that country which ranks second in the world in mortality from stomach cancer The investigators found that the three counties with the highest nitrogen fertilizer use during the period 1945-19h7a2d the highest adjusted gastric cancer mortality rates for the fifteen year period 1957-1971 These findings the authors state would agree with the demonstrated ability of certain bacteria to synthesize nitrosamines from secondary aromatic amines and nitrates within achlorhydric stomachs Liquid Parrafin | Boyd and Doll 56 determined the frequency of use of purgatives among 614 patients with gastrointestinal cancer and 1313 controls Of the 614 cancer patients 387 had cancers of the large bowel and 227 had stomach cancer Half of the controls had lung cancer and the remaining half had gastrointestinal diseases other than cancer Expected numbers of those who never occasionally or often used purgatives 7 were based on the experience of the entire group - both cases and JACKSON 025493 53 controls The overall difference between them in the use of purgatives after adjustment for age and sex was not statistically significant p 0.1 However when use of specific types of purgativweass considered it was found from an analysis of age and sex specific rates that the proportion of cases who used liquid parrafin was approximately twice as high as the proportion of users of this type among controls Analysis of the age and adjusted expected and observed numbers of often users revealed that among all gastrointestinal cancer patients combined and also among stomach and large bowel cases separately the use of liquid parrafin was statistically significantly more frequent among cases compared with controls Liquid parrafin is a mixture of liquid hydrocarbons obtained from petroleum While the purification process should remove most of the polycyclic compounds the authors found that this did not always occur They express the opinion however that liquid parrafin is probably not of direct etiologic importance but may be associated with some other unknown etiologic factor Graham et al 57 based on a control study of 188 male and 88 female cases of stomach cancer and 800 neoplastic digestive disease controls reported similar findings In their group cases were significantly more frequent users of cathartics although the specific types of cathartics were not determined These investigators caution that purgative use may be secondary to another relationship such as chronic constipation with this disease JACKSON 025494 54 Betel Chewing Schonland and Bradshaw 58 undertook an ecological study to determine the relation between upper alimentary tract cancers and betel chewing among South African Indians in Durban adjusted cancer incidence rates were calculated for Indians and English of both sexes and they found that Indian females had a significantly greater incidence of upper alimentary cancers than did English females while there was no differences between males Among the former the stomach cancer incidence was 100,000 while the latter group experienced an incidence of 12.4 100,000 The investigation further established that a far greater proportion of Indian women than men chew betel and that among chewers females chew more heavily than males Specifically 8.3 per cent of Indian males and 54.2 per cent of the females aged 20 and over are addicted to betel chewing No attempt however was made to analyze the cases themselves for betel chewing characteristics BATO JACKSON 025495 ANTECEDENT CONDITIONS A constellation of related factors - peptic ulcer gastric polyps gastritis achlorhydria and pernicious anemia ~ have been implicated in the etiology and pathogenesis of gastric cancer although the evidence for the involvement of these factors is equivocal and often in dispute Ulcers are only rarely if ever considered to be precancerous lesions although some investigators feel that they may be foci of increased susceptibility since the integrity of the gastric mucosa has been destroyed However it is difficult to distinguish an ulcerating primary carcinoma of the stomach from one developing secondarily in a peptic ulcer which complicates the problem of determining antecendent relationships Additionally an existent ulcer may possibly be overgrown and concealed by a secondary carcinoma thus masking the existence of both Numerous investigators have histologically examined ulcer series for the presence of cancerous development Elliot et al 59 examined 109 ulcers and found that 0.8 per cent in the greater curvature were malignant while 2.2 per cent of those in the prepyloric area were cancerous Cohn et al 60 differentiated peptic ulcers by size and found that of a total of 533 patients with ulcers 28 per cent of the ulcers measuring less than 2.5 cm in diameter were cancerous while 37 per cent of those greater than 4 cm were malignant JACKSON 025496 56 Hirohata and Kuratsune 61 investigated the geographical relationship between mortality from gastric ulcer and carcinoma in Japan where mortality rates from stomach cancer among males were the highest in the world in 1962-63 as were gastric ulcer mortality rates for both males and females Based on results of pathological examinations by other investigators the authors estimated the ulcer cancer rate to be between 2.5 per cent and 39 per cent and the relative risk of developing stomach cancer among those with ulcers has been estimated to be approximately five The investigators were of the opinion that the geographical variation in the incidence of gastric ulcer and carcinoma should follow similar patterns if either of two hypotheses are true the carcinoma develops from the ulcer or both conditions have similar etiologies Accordingly indirectly adjusted rates were calculated for two triennial periods 1949-51 and 1959-61 separately for each sex and each prefecture in Japan Correlation coefficients were then calculated for different combinations of calendar period sex and disease There was essentiallnyo correlation in either sex between mortality rateosf stomach cancer and ulcer during the two periods under study nor was there any significant correlation between the ulcer mortality rates during 1949-51 and the cancer mortality rates ten years later on the assumption that a substantial amount of time would be necessary for the development of cancer from ulcer The possibility that this latter comparison may be misleading because of the potentiality that those ulcer cases at high risk for some reason of developing cancer had already died leaving only the low risk ulcer cases alive was not discussed The authors do admit that ^"lcer JACKSON 025497 57 mortality rates are not the best measure of incidence because the case- fatality rate is low and that selective migration and accuracoyf diagnosis are potential problems but they do not feel that their conclusion of no association between the two diseases would be challenged by these shortcomings Gastric adenomatous polyps are also considered by some to be direct precursors of cancer Polyps are benign glandular neoplastic growths and cellular atypias are often but not essarily present . Bera 62 presents histologic observations on a series of 106 gastric adenomatous polyps from 45 patients Of these polyps 13 per cent showed a definite transition within the polyp from benign to invasive cancer whil6e5 per cent had some degree of cellular atypia Berg concludes that adenomatous polyps may be precancerous lesions and that they are reflections of generalized gastric neoplastic tendencies In attempting to demonstrate however that polyps are precancerous lesions it should be kept in mind that gastric polyps are very uncommon | lesions which are seen even less frequently than stomach cancer Peptic ulcers on the other hand affect as many as 10 per cent of the | population Ming and Goldman 63 examined 90 polyps removed from 49 patients 76 of which were classified as regenerative hyperplastic polyps ; | lacking distinct neoplastic nature Of the remaining 14 10 adenomatous polyps were considered to be truly neoplastic and in four of these 14 per cent focal malignant changes were demonstrated and these generally had marked hyperplasia and metaplasia Additionally there was a existing adenocarcinoma of six of the 49 patients JACKSON 025498 58 Tomasulo 64 studied 97 cases of gastric polyps and also differentiated between those which were hyperplastic and adenomatous Overall 34 35 per cent were found incidentally during gastrectomies for gastric carcinoma and five 5 per cent were incidental findings during gastrectomies for peptic ulcers and other diseases Overall 59 per cent of the adenomatous polyps and 28 per cent of those which were hyperplastic were associated with gastric cancer and changes identified as carcinoma in situ were seen in five of 23 cases of adenomatous polyps The authors conclude that adenomatous gastric polyps may be precursors of invasive gastric carcinoma and express the opinion that those who say they are not may be basing their evidence differentiated on an analysis which has not between the different | histological types Some investigators have also differentiated between polyps of different sizes Monaco et al 65 studied 153 patients 15 10 per cent of whom had large polyps greater than 2 cm in diameter which is generally considered the type most likely to undergo malignant transformation While there was marked epithelial atypicality and small foci of carcinoma in nine of these 115 cases none showed metastases and the authors conclude that there is no evidence in our series that adenomatous polyps progress to the usual invasive infiltrating carcinoma or that the residual stomachs are more prone to the development of carcinoma which disagrees with the conclusions of Berg 62 and Tomasulo 64 Marshak and Feldman 66 reviewed the literat^"reand concluded that estimates of the transformation of benign gastric polyps to a malignancy varied from 0-51 per cent and they felt that the disagreement may be due to differenceisn terminology and pathologic JACKSON 025499 criteria Based on available evidence they recommend that if on roentgenologic examination the polyp measures less than 2 cm in diameter conservative treatment should be undertaken If the polyp is smooth and greater than 2 cm in diameter however it is usually benign but since there may be such a finding with polypoid carcinoma, surgery is indicated Achlorhydria is probably important as a reflection of the integrity of the parietal cells in which hydrochloric acid is produced Gastric mucosal atrophy is therefore generally accompanied by a state of a- or chlorhydria There are some who believe however that gastric atrophy which leads to achlorhydria may actually increase susceptibility to carcinogens which might otherwise be destroyed by the normal pH in the stomach At any rate numerous studies have indicated that achlorhydria is in many cases concomitant with gastric atrophy and cancer For instance Comfort et al 67 determined that mean gastric secretory activity was lower at a mean interval of 11.2 years prior to the diagnosis of stomach cancer It was actually lower as early as the third decade of life but the decrement was more marked in the younger The the decades authors express the opinion that atrophy of ' gastric mucous membrane is responsible for this decreased gastric acidity and point out that gastric cancers develop most often in stomachs with lower than normal mean secretory activity many of them diseased with gastritis and peptic ulcers Stempien et al 68 obtained quantitative estimates of gastric acid secretion among patients with numerous conditions including super- ficial gastritis gastric ulcer and atrophic gastritis They found JACKSON 025500 60 however that only 4 per cent ofpatients with atrophic gastritis were hyposecretors compared with 17 per cent among those with normal mucosa and 28 per cent of the patients with gastric ulcers Surprisingly also their series indicated that among patients with ulcer and atrophic gastritis 28 per cent and 24 per cent respectively were hypersecretors. Pastor and associates 69 avoided the problems of sectional determinations and conducted two year retrospective cohort analyses on Japanese of both sexes In the first which included 7,000 people they determined that males with low serum pepsin levels experienced an adjusted stomach cancer incidence three times higher than those with normal secretion although this difference was not found among women The second study involved 1,250 Japanese and they found that in general persons with achlorhydria had a significantly higher adjusted incidence of gastric cancer than those with normal levels Additionally almost 700 patients had both tests and it was found that the highest incidence occurred among those with both low pepsin levels and achlorhydria Althougthhe site of production of intrinsic factor IF which is | necessary for absorbtion of Vitamin B12 without which red blood cells do not properly mature is presently unknown there is some recent evidence 70 that it is produced within parietal cells In the absence of IF pernicious anemia develops It would therefore not be unreasonable to find simultaneously achlorhydria and pernicious anemia in some gastric cancer patients especially when the gastric cancer has been preceded by chronic atrophic gastritis which is charaterized by a progressive and irreversible atrophy of the glandular epithelium of the stomach In addition atrophic gastritis is believed by some to be a percursor of gastric carcinoma BATO JACKSON 025501 6 Sivralla and Seppala 71 conducted a follow study of 53 patients with biopsy evidence of atrophic gastritis among whom there was no sign of gastric carcinoma on the basis of roentgenologic and gastroscopic examination As controls they chose 210 patients for whom gastroscopic examination revealed no signs of gastric carcinoma gastric ulcer or benign tumor and among whom the gastric mucosa appeared normal The controls and cases were very similar in sex and age distribution and in the follow period 5.8 years for cases and 6.0 years for controls Of the 52 cases a total of 20 either died during the observation period or refused examination with 33 undergoing examination at the end of the follow period Unfortunately the controls were not examined and the investigators apparently relied on the development of gastric carcinoma among the controls to somehow come to their attention Sign of gastric carcinoma appeared in five of the cases four of whom died during the follow period and one of whom was diagnosed as having gastric carcinoma at examination none of the controls showed signs of gastric cancer In only one instance however was the diagnosis of carcinoma confirmed by histologic examination while among the remainder the diagnosis was based on roentgenological gastroscopic or esophagoscopic examinations or upon macroscopic examination at autopsy Because of these limitations the authors hesitate to conclude definitely that atrophic gastritis is a precursor of the malignant lesion although they suggest that carcinoma might develop from advanced atrophic gastritis with metaplasia Kaplan and Rigler 72 examined a series of 43,021 consecutive autopsies at the University of Minnesota during the period 1915-43 for JACKSON 025502 62 the existence of pernicious anemia PA and gastric carcinoma The diagnosis ofpernicious anemia was based on departmental records when it had been made prior to death The investigators were concerned that the diagnosis of PA may have been discarded or unrecognizeidf a carcinoma was found so they essentially took a sample of autopsies to determine the proportion of occult cases Diagnosis of PA for this sample was based upon information collected about the patient during his lifetime since the best mortem evidence of the disease would come from an examination of the bone marrow which undergoes rapid mortem autolytic changes The results anemia 36 12.3 per cent also indicated that in 293 cases of had carcinoma of the stomach pernicious This proportion of stomach cancer was more than three times as great as the existence of carcinoma in the remaining autopsy population of similar age : +. | In a pair of companion articles 73 74 Morson first attempted to establish the presence of intestinal metaplasia in patients with duodenal and gastric ulcers and gastric carcinoamnad determined that intestinal metaplasia was most common in stomachs removed because of carcinoma 92 per cent compared with those removed for gastric ulcer 82 per cent and duodenal ulcer 65 per cent In cancer stomachs the metaplasia was fairly evenly distributed while it was most common at the pylorus in ulcer stomachs However the mean ages and presumably the age distributions differed in the three groups and while there was no adjustment for these differences Morson cites other studies ; where this was done and there was still more intestinal metaplasia in cancerous compared with cancerous stomachs In the second paper Morson demonstrates that carcinoma can arise from areas of intestinal metaplasia which was observed in 33 per cent of 107 gastrectomy sections JACKSON 025503 63 SUMMARY OF MAJOR FINDINGS International Rates great variability with highest rates in Japan Russia Eastern Europe and South American National Rates great substantial variability with highest rates in North Dakota and lowest in South Carolina Sex Ratio Female approximately 2.0 Social Class decreasing rates as social class increases Religion no major difference aside from differences in ethnicity and social class Migrant Studies first generation migrants generally have rates closer to home rather than host country but second and third _ generations gradually approach rates of host country Dietary Factors negative association with vegetable consumption positive association with highly salted foods and possible protective effect of antioxidants as well as possible association with nitrosamines Blood Type slightly higher among those with blood type A Alcohol no apparent association 10. Tobacco equivocal association findings weak association if any 11. Soil Type unexplained unexplained possible association but findings essentially unexplained 12. Occupational Studies inconclusive although may be elevated among metal products workers possibly associated with high dust levels JACKSON 025504 PATHOLOGICAL CONSIDERATIONS Histologic Type By far the most common histologic type of stomach cancer is adenocarcinoma implying involvement of the glandular epithelium and estimates of the proportion of all gastric cancers which are of this type range from 70-90 per cent Ming 75 has summarized the results of numerous surveys Table 11 and has determined that on the average adenocarcinomas account for 88 per cent of all stomach tumors whether benign or malignant followed by lymphomas 3 per cent and leiomyosarcomas 1.7 per cent In total these three tumors account for 93 per cent of all stomach tumors and almost 100 per cent of all malignant ones Stout 76 however who is probably the most frequently quoted authority on stomach cancer is of the opinion that the histological classification is hopeless and unrewarding since almost all stomach cancers involve mucous secreting glands or pits and the histological classification which presumably differentiates between different tumor types becomes very arbitrary Lauren 77 in 1965 citing Stout's dissatisfaction proposed a new classificatory system which he considered to be clinical ^fin nature According to this system there are two main types of gastric carcinoma - intestinal and diffuse - and classification of a tumor into one of these two types is based on general and cellular structure secretory properties mode of growth and clinical features A summary of the characteristics of intestinal and diffuse type gastric carcinomas is presented below JACKSON 025505 65 TUMORS OF THE STOMACH McNeer and Pack Marshall 9 Marshall Marshal Total Percent Combined Series Carcinoma . Carcinoid Leiomyosarcoma Rhabdomyosarcoma Lymphoma _ Plasmacytoma ~ Kaposi's disease Total , a . 470 5 24 1 39 2 1 542 oa - Benign Tumors or Tumor Lesions 310 3 13 1623 3 24 115567 67 16 3970 11 77 87.9 1.7 23 1 350 39 35 - 136 = 3.0 1 1613 4199 93.0 . Folyp Leiomyoma . Vascular tissue turrors 32 39 8 Nervous tissue turnees | Lipoma . ; _ Fibrous tissue tumors 4 4 Heterotopic pancreas 6 Eaterogenous Eaterogenous cyst . Inflammatory lesions mC 19 , Miscellaneous Total 109 20 16 1 2 1 ; 1 4 45 51 37 140 9 .. 4 W . +. 92 10 3.1 02 3 2 7 0.2 0.1 1 1 7 3 39 70 o4.. 3 E 0.1 82 Series Total 651 395 1768 1700 130.0 . Me eee The University of Maryland Hospital series not previously reported reported from 75 Jezi JACKSON 025506 66 1. Intestinal type distinct large glandular lumens with glandular structure rarely completely absent cells large and clearly defined with large nuclei and morphologic variability cells lining the glandular lumens are well polarized columnar cells few mucous secreting cells spread to a clearly and often distinctly defined area spread not wider in the mucosa than above the infiltrating part of the tumor ~ 2 Diffuse glandular lumens are rarely seen and cells are scattered cytoplasm indistinctly defined and cells are more uniform nuclei small and more regular in form cells lining the lumen are unpolarized and randomly grouped secreting cells very extensive wider spread inside the mucosa spread not well defined In a series of 1,344 patients Lauren found 53 per cent intestinal types 33 per cent diffuse and 14 per cent heterogeneous in composition He then correlated the two main types with sex age and other charac- teristics of the disease and found that 65 per cent of the intestinal type carcinomas and only 54 per cent of the diffuse occurred among males The mean age of patients with intestinal type was 55.4 while it wa quite a bit lower - 47.4 - for patients with diffuse type Both mean ages however are lower than the general estimates made by other investigators and Lauren did not further analyze mean age by sex The age distributions also differed with a greater proportion of diffuse patients below the age of 50 and a smaller proportion older than 60 The mucosa in the immediate vicinity of intestinal carcinomas was less frequently normal than among the diffuse types and a greater proportion of intestinal types 88 per cent exhibited gastritis and intestinal metaplasia in the surrounding mucosa and also intestinal metaplasia There was no correlation between the two main types and the site of the tumor within the stomach but the prognosis was slightly better among those with the intestinal type Lauren concludes that the two main types at least to some extent are caused by different aetiologic factors and that they differ pathogenetically as well JACKSON 025507 67 Munoz and Collelly 78 were interested in determining whether Lauren's two main types differed etiologically and whether the decline of gastric cancer rates within the U. S. was related to these types They slightly modified Lauren's classification and assigned 127 histologically confirmed gastric cancer cases from Hartford Hospital during two time periods - 1940-44 and 1960-64 - into one of three categories intestinal well differentiated adenocarcinoma diffuse undifferentiated adenocarcinoma and others In addition cases were classified by age sex and place of birth the latter to differentiate between high and low risk birth areas There was a total of 62 native and 65 foreign cases and 74 intestinal 45 diffuse and 8 others Intestinal type carcinoma was statistically significantly more common among those from high risk birth places Although the authors do not explicitly specify how risk was determined the foreign group included 44 Europeans and 15 Russians and Poles This difference persisted after controlling for sex and time period The age and sex distributions during the two time periods were similar and although there was a _ decrease in the proportion of intestinal types between the two periods under study the decrement was not statistically significant They also determined that the ratio of intestinal to diffuse increased with age in both sexes and in both risk groups and corroborated Lauren's finding that intestinal metaplasia was significantly more common among those with intestinal type carcinoma | Munoz and Collelly like Terris 33 were of the opinion that declining U. S. gastric cancer mortality rates could not be explained by the decreasing proportion of foreign in this country An JACKSON 025508 68 indirect adjustment was then carried out to determine morbidity rates by histologic type for the two periods although the method of calculation is somewhat unclear and no adjustment was made for nativity composition which would seem to be important in view of their finding that the intestinal type was more common among the foreign especially since the proportion of cases from high risk areas in each of the time periods is not presented Nevertheless the authors explain that the 38 per cent drop in gastric cancer incidence during the two periods is accounted for mostly by a drop in the intestinal type It seems questionable to calculate adjusted rates and draw inferences to the general population based only on their hospital sample from which a large proportion of apparent cases without histologic confirmation were excluded especially if there is some selective difficulty in making one or the other type of diagnosis The authors conclude that the decline in gastric cancer incidence ... is related to a reduction in the frequency of intestinal type carinoma Since this type is also related to intestinal metaplasia the etiology | of the latter should be investigated Munoz et al 79 conducted another study based on Lauren's classification to determine if there were differences between between high and low areas This one was based on 322 cases of gastric cancer from Columbia and Mexico and assignment to a high or low risk area was based on a single estimation by another investigator ofthe rates in the cities from which the cases were drawn While the mean age of the patients was similar in both risk areas neither the age nor the sex distributions are presented although male female ratios for both area and type are freely used The results indicate that a greater JACKSON 025509 69 proportion of cases from high compared with low risk areas were of the intestinal types 51.8 per cent vs. 35.9 per cent Location Estimates of the site of the tumor within the stomach vary slightly and these estimates are doubtlessly complicated by the fact that by the time the tumor is discovered and examineidt may be very large and probably involves more than one region Ming 75 has collected estimates from more than 7,000 examinations by different investigators and found the following mean ranges - pyloric 47-50 per cent lesser curvature 13-39 per cent cardia 10 per cent diffuse 6-8 per cent These findings generally agree with those observed by Stout 76 Figure 2 and it is clear that the most common site is the distal portion of the stomach Crane et al 80 for instance found that 75 per cent of the cases in their series of over 1000 occurred in the pyloric region | Metastases _ Determination of tumor spread depends naturally upon its site and stage although there are fairly clear patterns based on the data presented by Stout 76 Ming 75 and Crane 80 The most common . metastatic sites are - perigastric lymph nodes 80-90 per cent peritoneum omentum and mesentery 20-40 per cent liver cent pancreas 17-20 per cent lung 30 per cent 15-50 per Metastases to the stomach from other sites or direct invasions are very rare and estimates of their occurrence are one per cent or Tess JACKSON 025510 pylorus lesser curvature - cardia curvature Beko broken linein stomachstomach indicates junction 12 of antral and fundic mucosae - ne Lo 2 ; : . Bore ; Bt : . : - LOCATION OF 120 CARCINOMAS IN , IZO RESECTED STOMACHS ,; . the of gastric : Figure Figure 21. Two diagrama Mostrating the location in the stomach of $ consecutive cases of gastric ulcer and 120 120 cases of gastric carci igre 1 from Stat A. P. Tumans of the stomach Pull New York And Med 23 101-163 101-163 1397. K. from 76 JACKSON 025511 71 Prognosis The outlook for victims of stomach cancer is very bleak and prognosis is apparently not improving very much with time While survival rates depend to some degree upon age sex site extent of disease at diagnosis and treatment the picture is overall very dismal Based on data from the National Cancer Institute 2 the overall year relative survival rate adjusted for mortality among diseased people of similar age race and sex distributions is only 12 per cent The median survival time is approximately six months even if the patient is less than 45 years of age About half of all stomach cancer patients receive no treatment and among those who do surgery is the usual treatment This low treatment rate is due to the fact that half of all cases are diagnosed at a stage beyond localized or regional involvement The best prognosis is found among patients diagnosed in a localized stage and who are treated surgically among whom the year relative survival rate is 50 per cent although this group accounts for only about 12 per cent of all cases Similar survival patterns were found in England 81 JACKSON 025512 CARCINOGENESIS The field of carcinogenesis is very broad and encompasses many different areas such as the events which occur at the cellular cellular and molecular levels the histologic characteristics of malignant growth and the results of animal experimentation with specific carcinogens as well as their mode of action Unfortunately despite the tremendous efforts expended thus far little is definitively known about the carcinogenic process This very brief review will stress the broad concepts and attempt to establish the plausibility based mainly on animal studies of the existence of a chemical carcinogenic process within the rubber industry which may be responsible for the greater than expected number of cases of gastric cancer Histologic Characteristics There are innumerable definitions of neoplasia but perhaps the simplest and most appealing is one offered by Hopps 82 neoplasia is a new growth of tissue autonomous in nature serving no useful purpose obeying its own laws without repect for the organism as a whole and growing at the expense of rather than for the benefit of . _ the body This definition however describes the entire neoplasm The next question is what are the characteristics of a neoplastic cell compared with a normal one Robbins and Angell 83 have summarized these attributes and there seems to be general agreement among pathologists JACKSON 025513 73 concerning the behavioral characteristics of cancer cells loss of or decreased adhesiveness loss of contact inhibition increased motility increased contact guidance and synthesis of enzymes or metabolites which may be ability injurious to normal tissue and which increase penetrative . Normal adhesiveness between cells is dependent upon certain structural and electrical properties of cells Desmosomes normally anchor adjacent cells and contact is strengthened by anionic bonds between cells for which calcium binding by the cell is essential In neoplastic cells desmosomes are either absent or substantially reduced in number and this is couplebdy the inability of cancer cells to bind calcium which results in the presence of negative charges which repel each other on the outside of every cell decreased adhesiveness These alterations lead to One of the fundamental characteristics of normal cells is the inhibition of growth and mobility when they make contact with other cells Cancer cells on the other hand have not only lost this contact inhibition but there is evidence which indicates that mobility and growth are actually increased when they come in contact with normal cells Even in the absence of contact with normal cells however cancer cells have been shown to be more mobile and more capable of rapid migration than most normal cells with a few exceptions such as macrophages However the very mobile normal cells are generally not capable of mitotic divisions so the combination in cancer cells of increased mobility and retention of mitotic ability facilitates penetration and invasiveness Normal cells generally grow in an orderly fashion along strands of connective tissue in vivo or along grooves in a glass culture dish in JACKSON 025514 74 vitro grow in If cancer cells destroy these connective tissue strands they a disorganized mass If these strands are present however cancer cells will exhibit a greater than normal contact guidance which gives directionality to its mobility thereby increasing its invasive and penetrating capacity There is also fairly good evidence although total agreement is lacking in this aspect that cancer cells are capable' of synthesizing proteolytic substances such as hyaluronidase which destroys connective tissue substance in preparation for invasion by the cancer cells It has also been postulated that they synthesize toxins which are capable of destroying normal cells which enhances the propagation and spread of cancer cells Metastasis or spread of cancer cells is considered the most serious complication because surgical removal of a metastasized cancer even though it is not encapsulated as is a benign neoplasm generally results in a fairly good prognosis In the neighborhood of the primary growth metastasis generally is accomplished by direct invasion of adjoining tissue which is enhanced by the increased mobility and penetrance of cancer cells described above This often leads to penetration of blood or lymph vessels resulting in hematogenous or lymphatic metastases respectively The cancer cells then lodge in new tissue often far removed from the primary site which is called secondary foci or foci of metastases Once a cancer has metastasized the prognosis is drastically impaired Metastases are exacerbated by another feature of cancer cells namely their differentially high | reproductive rate which gives the appearance of being essentially autonomous in nature This increased rate of growth is especially| apparent in undifferentiated or anaplastic malignancies as opposed JACKSON 025515 75 to the differentiated type It should be remembered however that despite the rapid rate of growth of cancer cells there is a very long interval measured in decades between initiation of the process and clinical manifestations Chemical Carcinogens Before proceeding to a discussion of chemical carcinogenist would be best to briefly describe some terminology and alternative hypotheses of the carcinogenic process Direct acting or proximate carcinogens are those which do not require host enzymatic metabolism to generate the key reactive substance or ultimate carcinogen Many synthetiancd industrial chemicals are proximate carcinogens carcinogens however become active only after metabolic conversion by the host and are not carcinogenic in their native form Included in this group are most chemical carcinogens Proximate carcinogens are generally highly reactive substances which is almost a requirement of industrial chemicals since they are normally quickly deactivated If they are not deactivated however they become highly carcinogenic carcinogens can also be either activated or inactivated a process which generally occurs in the liver which contains the enzyme systems necessary for metabolic conversion In fact it is believed that all carcinogens can be activated by the liver and also by other organs such as the lung kidneys and placenta among others Note however that proximate carcinogens do not require activation while carcinogens do although both may be deactivated There is another important general concept of the difference between initiators and promoters An initiator is capable of initiating JACKSON 025516 76 the carcinogenic process by itself and is therefore a true carcinogen but the process is often very slow Carcinogensis can be hastened however by application of a promoter or carcinogen which is itself generally not carcinogenic Application to a site of a promoter before the addition of the initiator generally does not result in cancer unless the interval is very short in which case the promoter may still be active This hypothesis is called the Stage Theory of Cancer Induction although it should be remembered that the initiator by itself is capable of inducing carcinogenesis in large enough doses Therefore while the initiator may or may not require a promoter promoters can do no damage in the absence of initiators since they are not independently carcinogenic Regardless of the pathway or mode of action of a specific carcinogen the cellular alteration must be heritable This alteration then becomes capable of storage replication and transfer to all new cells and indeed new generations of cells may incorporate additional changes which also become heritable and further increase the difference between normal and neoplastic cells The mechanism of action of alteration however is presently moot with theorists divided between two alternative pathways genetic or epigenetic Proponents of the genetic theory insist that cancer induction results from an interaction of the carcinogen with DNA resulting in a permanent alteration in the DNA information ' which naturally becomes heritable This is generally believed to occur by a binding of the carcinogen to nucleic acids resulting in a permanent alteration of the base sequence According to this theory initiators are capable of inducing just a few mutations which are not capable of autonomous growth or replication but the application of a JACKSON 025517 77 promoter which may lead to additional alterations in the presence of the promoter may lead to heritability and autonomy The epigeneticists on the other hand believe that primary mechanism involves some alteration in the cytoplasmic feedback mechanism perhaps by some alteration of messenger RNA which may lead to a permanently altered but stable metabolic pathway According to this theory the carcinogenic process is reversible because the basic alteration is in the gene expression and not in the genome itself In a classic review article Miller 84 who is one of the pioneers in the field of carcinogenesis has reviewed the pertinent findings to date He notes that almost all chemical carcinogens with the exception of alkylating agents such as uracil mustard are carcinogens which require activation before they become ultimate carcinogens The single basic characteristic of almost all ultimate carcinogens whether or not they require activation is their strong electrophilic property which enables them to react with nucleoplilic substances such as nucleic acids and proteins For instance it has been shown in vivo and in vitro that the aliphatic agents including nitrosamines are alkylating agents either in their native form proximate carcinogens or after enzymatic or enzymatic activation carcinogens which interact with nucleic acids and protein Aromatic carcinogenosn the other hand have not been shown to bind with nucleic acid in vivo but the evidence is very convincing that they form strong electrophilic reactants which can displace at least in vitro the weaker electrophiles such as + Despite the magnitude of the effort invested in elucidating the mechanisomf chemical carcinogenesis however Miller concludes that in case of chemical carcinogenesis do know unequivocally the nature of the JACKSON 025518 molecular target critical to the induction of that carcinogenic process The known carcinogens of greatest interest within the rubber industry are the aromatic hydrocarbons including 3,4 benzpyrene nitrosamines and asbestos The remainder of this discussion will focus on these agents It is generally believed that the aromatic hydrocarboanrse precarcinogens which require activation Kellerman et al 85 have shown that the ultimate carcinogen is probably an epoxide which is formed in the presence of the inducible enzyme aryl hydrocarbon hydroxylase of numerous AHH which shows increased activity after administration chemical agents including benzpyrene They have further demonstrated that the U. S. white population can be divided into three phenotypic inducibility groups - high intermediate and of inducibility is presumably under genetic control with low The degree two alleles at a single locus resulting in homozygous high and low groups and the heterozygous intermediates They then determined AHH activity in cultured leukocytferosm three different groups ~ histologically confirmed cases of bronchogenic carcinoma cases with a range of different types of tumors and a healthy control group While no mention is made of confounding factor distributions in the three groups this presumably is not an important factor in this type of study They found highly . statistically significant differences in the phenotypic distribution between lung cancer cases and healthy controls 30 per cent of the cases had high inducibility compared with 9.4 per cent of the controls while only 4 per cent of lung cancer cases had low inducibility against 45 per cent of the controls The differences between the other tumor group and healthy controls were not significantly different They JACKSON 025519 79 conclude that this evidence supports the hypothesis that susceptibility to lung cancer is inherited Unfortunately these results have not been duplicated after attempts by numerous investigators There is fairly good evidence the polycyclic hydrocarbons require suspension in particulate matter to be carcinogenic Winkelstein and Kantor 86 have studied the relationship between suspended particulate air pollution and gastric cancer in Buffalo New York among males and females 50-69 years of age According to the authors the finding that mortality rates from gastric cancer were twice as great among residents of high compared with low areas of suspended air pollution is independent of socioeconomic and ethnic effects To eliminate effects of a large Polish population the investigators assumed that this group would have a rate 100,20 -10ti0 mes as high as the average death rate among whites in the age range under study then from a census estimate of the proportion of the total population of Polish origin they calculated the number of deaths expected among this ethnic group These deaths were subsequently removed from the numerators and denominators which were used to estimate mortality experience This appears to be a novel approach to the problem which may or may not be methodologically sound but their estimate that a mortality rate of 100,000 100,000 is 2-1 times the average is apparently not corroborated by the evidence which they themselves present indicating that among males as the component spectrum of pollution levels from low to high is traversed the rates are approximately 25 40 50 and 100,000 At any rate the evidence indicates that polycyclic hydrocarbons such as benzpyrene must be suspended in particulate matter to be JACKSON 025520 80 carcinogenic Both the carcinogenic hydrocarbons and particulate matter in the form of carbon black and perhaps talc are abundantly present within the rubber industry Matolo et al 87 studied the incidence of gastric cancer in two coal mining counties in Utah and compared these rates with those foundin one coal mining county and with the entire state They found that the incidence of gastric cancer in coal miners was 15 times higher than in coal miners in the mining counties and 40 times higher than in miners in the entire state of Utah The authors stress the fact that coal soot has a rather high benzpyrene content and contains both large and small particles to which this hydrocarbon may adhere They conclude that there is a statistically significantly higher incidence of gastric cancer in the two mining counties and add that coal mining and the extensive use of soft coal by both miners and miners for heating and cooking might be an etiologic factor Creagan et al 88 however suggest that the correlation may be with socioeconomic class rather than with exposure to coal mining and its use In a study on the relationship between atmospheric pollution and mortality from certain chronic diseases in England and Wales however Percy Stocks 89 found that intestinal and gastric cancers were related significantly with smoke containing many different types of polycyclic hydrocarbons Stocks further states that this relationship can only be partly explained by social differences in the populations concerned There has been much research conducted on the carcinogenicitoyf many different types of nitrosamines since 1956 when Magee and Barnes 90 discovered the development of malignant liver tumors in rats fed JACKSON 025521 81 dimethylnitrosamines In subsequent animal investigations these compounds have been shown to be very potent carcinogens at many dif- ferent sites including the trachea lung kidney bladder and stomach in many mammalian species fish and birds Nitrosamines have been shown to be mutagenic and teratogenic as well as oncogenic * It has been reported that nitrosamines can be formed in vivo in the stomach via an catalysed reaction between a secondary amine and nitrite and also by a catalysed reaction at site in the presence of a secondary amine and nitratoer nitrite Hill et al 91 studied mortality from gastric cancer in ten towns in England - one experimental town with a mean nitrate content in drinking water of 90 liters and 9 control towns with a mean content of about 10 mg liters While mortality from all malignant neoplasms combined was significantly lower than expected mortality from stomach cancer was significantly higher for both sexes taken together with the greatest _ excess in females The authors hypothesize the formation of nitrosamines in the bladder which subsequently induces gastric cancer via a systemic reaction A number of investigators 92,93 have expressed the opinion that industrial workers are at particularly high risk of exposure either to nitrosamines or their precursors Numerous studies have shown that pleural mesothelioma and other pulmonary diseases are statistically correlated with asbestos exposure Recently a number of studies have suggested that gastric carcinoma as well may be associated with exposure to asbestos and asbestos- containing talc 94-96 131-139 and talc is widely used during tire production Industrial hygienists at the Occupational Health Studies Group have estimated that the asbestos content of talc used in the JACKSON 025522 82 industry may have been fairly high in the past Merliss 94 has suggested that the high incidence of gastric cancer in Japan may be due to treatment of rice with talc although others believe that the increased Japanese incidence may be due to the formation of nitrosamines as described above In study of talc miners in New York State Kleinfeld et al 95 found a significantly higher than expected mortality from gastrointestinal malignancies Other studies however disagree with the conclusion that gastric carcinoma is associated with exposure to asbestos In an interesting study which bears on these matters an English group 96 used an analytical electron microscope to study the particu- late material present in stomach tumors in Japanese Their analysis indicated the presence of Kaolin aluminum and calcium silicates talc and some types of asbestos fibers including both amphibole and chrysolite Unfortunately the authors do not mention anything concerning the levels of these substances in control tissues nor do they discuss the implications of these findings or their relationship to carcinogenesis JACKSON 025523 REVIEW OF MAJOR OCCUPATIONAL STUDIES Awareness of occupationally related diseases reaches far back into antiquity and scientific investigations to determine their causes have been conducted with varying degrees of success for hundreds of years Included in this chapter will be a review of the methodologies used to determine unique disease occurrences in three occupational groups - steelworkers asbestos workers and uranium miners These have been chosen because of the thorough manner in which the studies have been conducted and because it was felt that they are fairly representative of the methods generally used in occupational studies The purpose of this short chapter it should be kept in mind is not to present results or conclusions but rather to illustrate the methodologies generally employed by investigators to establish relationships between disease and exposure in occupational groups Steelworkers These studies conducted by a group from the University of Pittsburgh Graduate School of Public Health in collaboration with the U. S. Public Health Service and the National Cancer Institute are both recent and exhaustive and are very similar to our own rubber study within which this present investigation is being conducted For this reason all papers thus far published will be reviewed in detail The introductory paper 97 begins with a general description of the methodologic approaches used in long industrial studies and includes an historical review The authors caution the reader about an JACKSON 025524 84 ignorance of selective factors of major concern in studies of this nature which may lead to biased estimates of disease occurrence and also describe additional sources of bias such as improper definition of the population or an inappropriate choice of comparisoonr referent group As an illustration of the latter the authors use the example of what we call the healthy worker effect which may lead to incorrect conclusions about the health status of an employed population To overcome these types of problems they recommend longitudinal studies although they note the impracticality of this approach for - diseases with long latent periods a category which encompasses essentially all chronic diseases and not only malignancies They suggest instead and themselves use the retrospective cohort approach The cohort used for this study includes all men employed by three firms in 1953 in seven plants within Alleghany County Pennsylvania This cohort represents approximately 60 per cent of all men employed by the steel industry within the U. S. in 1953. In all just under 60,000 steelworkers were included and follow information on all but 97 0.2 per cent workers was obtained with all deaths confirmed by death certificates Since analyses according to specific occupational history would routinelbye performed throughout the course of the study a method was needed which would assign each member of the cohort to his appropriate occupational group As in our study changes in job title terminology over the years as well as different usage by each of the three firms involved made this task more difficult Accordingly personnel offices at each plant were requested to provide a translation to currently used titles and in this manner the 14,000 distinct titles JACKSON 025525 85 were reduced to 76 subgroups and further into 10 basic groups a procedure very similar to the one utilized by the rubber study Initial simple mortality analyses were based on SMR's with expected numbers calculated from county rates separately for whites and blacks Interestingly the first more detailed mortality analyses 98 were correlated with economic rather than physical factors In 1947 an income index used to determine wages was introduced into the industry Each job according to this system is assigned a total score and a corresponding hourly wage The total score comprises twelve elements which assess skill responsibility effort and working conditions on the job Because this total job score is the industry basis for the determination of salary level it concomitantly measures income and reflects the education and work experience of the employee Use of this total score avoids the problems caused by fluctuating conditions in that it remains relatively stable and is therefore economic a more reliable indicator of socioeconomic status than are the actual wages These scores were further categorized for analytic purposes into one of four groups In general a negative association was found between total job score level and mortality among whites Additionally the earlier reported excess white mortality was explained by job score level and racial factors were therefore ruled out an instance where the investigators did not stop at the holy duality because all subjects in this study are male but sought real explanations other than the nebulous factors of race and sex for their findings The next paper 99 in this series describes in detail the methods used for tracing large industrial populations over extended JACKSON 025526 86 periods of time The authors stress the importance of determining the vital status of as large a proportion of the cohort as possible because of the possibility of drawing erroneous conclusions or biased estimates of disease occurrence They especially caution against relying on data only for those who remain in an industry in which case total mortality will be understated Additionally distribution of cause of death will likely be different for those who remain withiann industry compared with those who leave The investigators examine the operation of some of these factors within the present study where 1381 of 4716 deaths at the time of this analysis among the cohort members were unknown to the employers The differential distribution of the unknown causes of death was striking -- 45 per of deat deh ats hs from malignancies compared with 21 per cent of heart and circulatory disease deaths were unknown to the employers There was a racial difference as well with employers unaware of 28 per cent of the deaths among whites and 36 per cent of deaths among whites They also indicate that employee age is itself a selective factor which may lead to biased estimates because the pension system favors greater awareness among employers of deaths among older compared with younger employees The authors then review available sources of follow information including the Social Security Administration Veterans Administration county death lists city directories and Internal Revenue Service sources among others The follow scheme is summarized in Figure 3 and Tables 12 and 13 JACKSON 025527 87 Redmond et d ~ from 99 Follow scheme for tracing steelworkers steelworkers JACKSON 025528 TABLE 12 Redmond et ol Distribution of Steelworkers by Primary Source of Follow and Place Primary Scurte ; Toul Employee Record at Flant County Death Lists 1953-1962 P. Palk & Telephone Directories Post Office Internal Revenue Service : Social Security Administration : Telephora Contacts Board of Public Assistance Burau of Employment Security Eureau of Missing Persans , Other Sourcza Not Located * Indicates valves less than 0.05 . Total No. 6 White No. % Nonwhite No. % 24,970 3,500 100.0 19.0 21,437 2,855 100.0 13.3- 13.3- 3,483 645 100.0 15.5 7.137 4.6 943 4.4 191 5.5 3.557 14.2 9.141 30.6 3,203 14.9 8,097 37.7 349 120 1,04 320 2,828 4,043 504 56 20 11 11.3 2,516 11.7 312 16.2 = 3,309 15.4 1 734 2.0 11 383 1.8 7 121 7 0.2 11 20 0.1 2 0.1 22 0.1 7 . 0.0 & 0.0 3 9.0 21.1 3.5 lag O23 O14 . 67 0.3 50 0.2 17 97 0.4 70 0.3 27 ag from 99 88 JACKSON 025529 TABLE 13 Redmond et al Primary Source Distribution of Steelworkers by Primary Soume of Up Follow and Year of Separation 1953-1955 1956-1953 No. % No. % Total 11,805 100.0 5.292 100.0 Employee Record at plant County Death Lists 1953-1953 R. Polk 2 Telephone Directories Fost Office 91 8.4 472 4.0 856 7.3 3,906 33.9 20 12.9 377 7.1 985 18.2 1,981 37.4 Internal Revenue Service Social Security Administration Telephone Contes Board of Public Assistance 1,994 2,959 342 42 16.9 25.1 2.9 - 0.4 504 9.5 658 12.4 21 1.7 9 0.2 Eureau of Employment Security Bureau of Missing Persons Other Sources 20 0.2 S 0.1 42 0.4 2 0.0 3 0.1 11 0.2 Not Located 24 0.7 11 0.2 1959-1862 No. 4, 7,372 100.0 1.929 23.2 223 2.7 1,738 22.1 3.164 40.2 330 4.2 428 5.4 31 0.0 5 0.1 7 0.1 7 0 7. 14 0.2 2 0.0 Indicates values less than 0.05 from 99 . 4 4 89 JACKSON 025530 90 In the fourth paper of this series 100 the authors describe the methods used to determine mortality by work area a procedure which leads to the identification of general and specific high and low mortality risk areas within the plant when compared with the . experience of the total steelworker population The authors again caution us about the attribution of unusual disease occurrences within specific work areas to environmental sources of etiologic associations exposures without due to bias considering ; For instance in using the total steelworker group as a referent . selection due to employability healthy worker effect is avoided but selection for health within a specific work area is not and as a consequence deficits or excesses within an area may be unrelated to exposures Because little is known about these internal selective factors much caution is suggested before drawing any conclusions of an etiologic nature In all 53 work areas were defined and workers were assigned to one of two groups - work area at initial observation in 1953 and areas in which at least five years had been spent by the end of 1961. Expected deaths within each work area were race - , - age , nativity - , and residence and were calculated on the basis of rates from the total steelworker population which were then compared with observed rates within each area Analyses were_first performed according to the categorization by . work area in 1953 and these were followed by analyses of workers with a minimum of five years of experience within a specified work area This second group was included because of the probability that disease response is a function of duration of exposure in which case the first JACKSON 025531 91 type of analysis by including men with very short exposure durations may mask true differences In addition because it was assumed that the mortality experiences of workers with long and short employment durations differed the comparison group for those with five years of experience within an OT was comprised of steelworkers who had attained five years of employment within the industry prior to or during the period 1953-61 Next 101 an earlier finding of a fold excess of respiratory cancer among men employed in the coke plant is examined and a differ- ential mortality analysis among whites and whites separately by calendar period and by duration of employment in several specific areas within the coke plant is performed Once again an internal comparison is made between the observed mortality experience during the period 1953-61 and expected mortality predicted by the age - ' race - ' and nativity rates for the total steelworker population within each of two work areas within the coke plant - coke oven workers and nonoven workers Also two different employment groups were examined - those employed in the coke plant in 1953 and those not employed in 1953 but with experience in the coke plant prior to that date The latter group was included so as not to restrict subjects to those employed at a single point in time among whom certain selective factors might be operating In all 3530 the excess respiratory cancer was men were studied and it limited to men employed was found that at the coke ovens When this was further analyzed it was found that among those working at the tops of the ovens there was a fold lung cancer risk and of these among those with five or more years of experience the relative risk was ten Again an earlier reported white difference JACKSON 025532 92 was explained by differences in work histories which presumably indicates an environmental and not a racial difference This finding among Allegheny county steelworkers led to an additional study 102 in ten selected plants within the U. S. and Canada in an attempt to establish confirmatory evidence for the excess risk reported above The study subjects included men in these plants with coke oven experience during the period 1951-55 and the experience of this group was compared with the experience of a group of men employed at the same plant during the same period Two controls were chosen for every case matched on race and date of initial employment and while two different matching schemes were used the basic procedure was the same at all plants The analyses utilized various work areas within the coke oven area full topside partial topside side oven as well as varying employment durations and contrasted the rates in these different groups with those of men employed in different work areas of the plant with adjustments for specific plant race age at entry and calendar years of follow Ingeneral the conclusions reported earlier were confirmed by these studies which also yielded a new finding ofa kidney cancer excess among this group The next three papers in this series 103-105 examine disease occurrences within three specific groups of workers - crane operators open hearth workers and sheet and mill workers In each instance an appropriately chosen control group from within the industry was used for the estimation of relative risks Cranemen were classified according to their most likely type of exposure gaseous particulate a gaseous and particulate clean unknown and were compared with a group JACKSON 025533 93 of controls who were never cranemen In addition exposure duration estimates were utilized It was found that men with particulate exposure had higher renal and cardiovascular mortality risks than men in other categories and there was a generally increasing relative risk as exposure duration increased with the expected exception of those who were employed in relatively clean area The last two studies were conducted in ways very similar to those already describaendd need no further comment especially since neither presents definitive conclusions Similarities between the steelworker and rubber worker have been noted throughout this review although there is one major difference in the approach to matched studies We generally general y identify individuals with a specific disease and compare them with appropriately chosen controls in the classic control mode during which we look for differences in work histories which would indicate different environmental exposures and from which we estimate relative risks The steelworker group has chosen the follow retrospective cohort approach in which their two groups are exposed - not exposed as of a certain date from which more precise and direct relative risks can be calculated Because our initial analyses were not performed performed by work area OT and because our initial identification of mortality excesses and deficits were not calculated according to work area this approach would be more difficult within the rubber study Asbestos Workers Lynch and Smith 106 in 1935 were the first to report the case of an asbestos worker with both pulmonary asbestosis and associated JACKSON 025534 94 carcinoma of the lung and they speculated on the possible etiologic significance of asbestos exposure in pulmonary malignancies Since this first report many others have found the joint occurrences of these two conditions In 1960 Wagner et al 107 identified 33 cases of | pleural mesothelioma a generally rare tumor in an asbestos mining area in South Africa among both whites and blacks of both sexes Of the 33 cases the investigators were able to establish definite evidence of asbestos exposure among eight cases anda probable exposure among another twenty Numerous more thorough studies followed these initial reports of the relationship between an occupational exposure and a malignancy Doll 108 in 1955 cited autopsy data from England on persons known to have been employed at a large unidentified asbestos factory and showed that of 18 cases of lung cancer among this group 15 were associated with asbestosis He then estimated the latency period among these 15 men by determining the elapsed time between first exposure and death which varied from 16-36 years among those with ~ associated asbestosis while the exposure ranged only from 2-11 years among the three cases of lung cancer without asbestosis All 15 men in whom both conditions prevailed had started work in the industry before 1921 and therefore had at least ten years of concentrated exposure before national standards were instituted in 1933 Doll also used a prospective approach in his investigation reported in the same paper of 113 men employed for at least 20 years within the industry in areas of high asbestos exposure based on personnel information supplied by the company under study The group included some who were still active and others who were no longer JACKSON 025535 95 employed within the industry The group was followed and causes of death were determined from death certificate information This was followed by the calculation of an indirectly standardized rate using the male rates for England and Wales as a referent A total of eleven deaths due to lung cancer with mention of asbestosis was found while none were expected From these data Doll concluded that asbestos workers experience a lung cancer risk ten times higher than the general male population and that the risk became progressively lower as the duration of employment under dusty conditions decreased Selikoff Selikoff 109 who has studied asbestos workers probably more exhaustively than any other investigator studied and examined asbestos insulation workers in the New York - New Jersey area where the first union was established in the late 1800's He identified all 1522 men who were ever members of the union local under study during the period 1942-62 of whom 1258 were still alive at the start of the study in 1963. Over 90 per cent of this group which included actives retirees and withdrawals from the industry presented themselves themselves for a history and physical examination Detailed work histories were solicited from the men- The information included exact type of work performed use of protective respirators prior health experiences and smoking habits This was supplemented with a physical examination which included rays Overall more than half of those examined had evidence of pulmonary asbestosis and the proportion with this condition increased with increasing exposure the latter having been categorized into five exposure levels In addition eleven cases of lung cancer were discovered during these examinations all among men with 20 or more years of exposure within the industry JACKSON 025536 96 - Concentrating on the 632 men with 20 or more years of exposure among this group Selikoff in a subsequent analysis 110 calculated SMRS based on U. S. rates Among the group under study almost half 339 had 40 or more years of exposure and in all the group had experienced 8738 years of exposure within the industry SMRs of 700 for cancer of bronchus and pleura and 300 for stomach and colorectal cancers were reported although the latter were based on small numbers _ Enterline 111 identified 2833 white males who had worked in the asbestos products mainly textiles industry between 1948-51 for 13 different manufacturers using social security records Deaths were identified from death claims made to the Social Security Administration and of the 286 deaths so identified certificates were obtained for 273 The mortality experience of these asbestos workers was compared with the experience of cotton textile workers because the latter were thought to be engaged in work similar to that of asbestos workers and to have similar economic status without concommitant exposure In all 6243 cotton workers were included and specific SMRs were pr^'sented The causes SMR was 112 for asbestos workers and 88 for cotton workers while the respiratory cancer SMR was approximately 200 among the former and 30 25.6 among the latter Braun and Truan 112 undertook a cohort study after complaining about the absence in prior studies of a well defined population of asbestos workers and improper epidemiologic methodology They identified identified an apparently exhaustive cohort of asbestos workers in Quebec all of whom had at least five years of experience as of 1950. The cohort JACKSON 025537 97 | included 6091 men 133 of whom were lost to follow with the remainder being followed for six years During this period there were 187 deaths of which nine were definite and three highly probable lung cancers Members of the cohort were studied with respect to age duration of employment smoking habits and a weighted exposure average based on number of years spent at various dustiness levels These exposure categories were determined from consultations with persons familiar with the environment and conditions in the various work areas Three levels were used with category II considered to be twice as dusty and category III three times as dusty as category , quite an assumption in view of the manner in which dustiness determinations were made While complaining about the paucity of epidemiologic expertise in earlier studies the authors themselves employ questionable analytic techniques and epidemiologic reasoning on numerous occasions and conclude that asbestos miners in Quebec do not have higher death rates from lung cancer than expected and further that the mortality rate does not increase with increasing length or severity of exposure Selikoff 119 concerned that most studies do not differentiate between the different types of asbestos such as chrysolite crocidolite anthophyllite and amosite identified all workers in an unnamed company who started work between 1941 when production began and 1945 From a review with plant management from product samples and from workmen's old respirator filters it was ascertained that only amosite was used in this particular plant which continued production until 1954 In all 868 or 93 per cent of the original group was traced and expected deaths for the period 1960-71 were determined Selikoff calculated JACKSON 025538 98 2.4 expected deaths from lung cancer while 25 were observed although smoking histories since unavailable were disregarded Meurman et al 114 studied anthophyllite miners in Finland who had worked in one of two mines for three months or more between 1936-67 with the roster supplied supplied by the company Location and follow of subjects was accomplished by use of the Finnish national population registry complete and for the detailed records of 95 per cent of the more the state than 1000 church Follow was employees who met the inclusion criteria Interestingly a comparison group was selected from a nearby nonmining and nonindustrial town and every asbestos miner was matched with a control for date of birth sex and date of death if the miner was deceased In addition a questionnaire was sent to each miner and control soliciting information on current health status smoking habits and the duration qualitanyd intensity of work in a dusty environment as well as age at first exposure In all there were 248 deaths with cases exhibiting a lung cancer excess compared with controls while digestive system cancers were found in | equal frequency in the two groups McDonald 115 in a large study investigated the mortality experience of chrysolite miners in Quebec Records were collected from 44 companies listing all current and previous employees as of 1966 Approximately 28,000 employed as of 1966. miners were identified a Detailed work histories quarter of whom were including more than 13,000 different jobs were collected This number was reduced to just under 6000 and for each of these and for every year of operation average dust exposures were assessed on a point scale based partly on environmental measures conducted since the late 1940's and for JACKSON 025539 99 levels prior to that date based on interviews with employees of long standing Of the original 28,000 members a cohort of about 12,000 was selected comprised of persons born between 1891-1920 provided that they had been employed for at least one month and a detailed search was conducted for all lung cancer deaths among this group Observed and expected mortality rates were calculated and in the highest dust category in which were found 5 per cent of all subjects the agestandardized mortality rate was 20 per cent higher than in other groups In addition the difference in respiratory cancer mortality rates between individuals with minimal and maximal dust exposure was fold Lung cancer is not the only malignancy associated with asbestos exposure In two separate hospital control studies 116 117 neither of which can properly be called an industrial study since they were not conducted within specific industries but both of which are quite interesting a relationship between asbestos exposure and laryngeal cancer has been indicated The first was conducted in England on 100 consecutive laryngeal cases matched with malignant controls where it was found that 31 per cent of the cases and only 3 per cent of the controls reported industrial asbestos exposure with an average of 27 years exposure The second included 43 cases in Canada each matched with one control for age sex and residence within metropolitan Toronto From personal interviews occupational exposures to a number of environmental agents including asbestos were determined and smoking histories were obtained Among the cases ten had prior industrial exposure to asbestos while none of the controls reported similar exposure The mean latent period JACKSON 025540 100 was 28.2 years range = 1-55 although these estimates are based on rather small numbers An interesting finding was that among men exposed at a younger age the latent period was longer although this " is not discussed in any detail The two studies therefore report similar findings regarding latency period and the proportion of all cases with an industrial asbestos exposure approximately 25 per cent which large seems highly significant especially if replicated in other ny studies which utilize more detailed work histories Uranium Miners Lorenz 118 in a 1944 paper summarizes the history of the elucidation of an association between uranium mining and pulmonary diseases Among miners at Schneeberg Germany and Joachimsthal Czechoslovakia in the Erzgebirge or Ote Mountain region between . Saxony and Bohemia an ailment called mountain disease or Bergkrankheit was described described as early as the sixteenth century In 1913 German investigator using a proportional mortality approach reported that malignancies of the lung accounted for approximately 40 per cent of all deaths among miners in this region He further stated that during the period 1875-1912 there had been among this group of miners 276 deaths from lung cancer 64 from tuberculosis 119 from other lung diseases and 206 from other causes Over the years investigators in this region continued reporting proportional | mortality lung cancer rates of 40-70 per cent among miners Included in this fine paper are a summary of the physical findings .o regarding radium exposure in the mines a review of the data up to that point from animal exposure studies as well as a rather depressing JACKSON 025541 107 description of the stark conditions in which the miners worked and lived In addition a summary is presented of the different etiologies to which the excess lung disease was attributed over the years including fungi mud and dust containing arsenic cobalt and silica Radon was first suggested as an etiologic agent in 1924 Archer 119 describes the initiation of studies of uranium miners in 1950 based on the knowledge that uranium and its decay products radon daughters ) are toxic and based on the evidence from Schneeberg and Joachimsthal cited above Health examinations were initiated in 1950 and continued through 1960 and included a physical exam and laboratory procedures as well as obtainment of an occupational history and pertinent social data Preliminary analysis indicated twice as many observed as expected cases of lung cancer although based on very small numbers 5 expected as well as an excess cardiovascular disease and accident rate In a rather large study of underground metal miners Wagoner et al 120 selected miners who had completed at least 15 years of employment | _ between 1937-48 The year 1937 was chosen as the initial date because the Social Security System was then inaugurated and it was used as a source of follow information Mortality analyses were performed through 1959 Of 75,000 employee company records reviewed 1759 miners met estimates the study criteria Because of discontinuous work patterns J of the number of years of exposure were based on the number of under- ground shifts worked and for analytical purposes the number of person of exposure was used Follow was accomplished through the use of insurance records state vital statistics departments x company retirement programs and social security records BATO JACKSON 025542 102 In all death certificatfeosr 930 miners reported to have died during the study period were obtained All other members of the cohort were assumed to be alive a procedure which makes conservative the estimates of excess mortality Overall mortality and cause distributions among miners were compared with similar findings in the state in which the miners worked although the name of the company and its location were not divulged In this manner expected number of deaths for each cause and calender year were calculated The study population was divided into two groups Cohorts I and II as follows those in Cohort I must have completed at least their fifteenth year of underground mining before alive as of that date members of Cohort II 1/1/37 and must completed their have been fifteenth year between 1/1/37 and 12/31/48 These groups were created to determine - whether survivorship bias existed among the members of Cohort I as would happen if the men who met the fifteen year requirement before January , 1937 and survived to enter the study were hardier than those who completed their fifteenth year at a later date specific mortality ratios were computed separately for each cohort and excess was essentially limited to diseases of the heart and lungs The excess in lung cancer SMR = 300 was not attributed to excess cigarette smoking among underground metal miners on the assumption that their smoking histories would not differ substantially from the histories of uranium miners among whom lung cancer mortality was increased only 10 per cent due to smoking habits Nativity effects were controlled by separate analyses for both foreign and native and the resultant SMRs were elevated in both groups although the excess was greater among native JACKSON 025543 103 Autopsies had been performed on 15 per cent of the subjects who died according to death certificates and verification by either pathological or radiographic examination was available for 73 per cent of the deaths Since these figures compare and verification rates among similar groups favorably with autopsy from the general population the lung cancer excess is not ascribed to differences in death certifi- cation An examination of the data presented in this paper indicates that on the average employment within each of the two cohorts lasted 15 years 25,033 person of exposure for 1759 persons although one must wonder why there were not more person of exposure since 15 years was the minimum inclusion criterion Also while the authors postulate radioactive inhalants as being of possible etiologic importance no measure of their prevalence within the industry are reported Wagoner et al 121 conducted a major study of uranium miners in the Colorado plateau with the intention of conclusively demonstrating that the excess lung cancer among uranium miners is due to airborne radiation an association which was demonstrated by Archer in 1962 The pool of available subjects came from 5370 miners in the Colorado Plateau a group which was the basis for numerous additional studies soon to be described Of this group 3415 white uranium miners had earlier volunteered for a physical examination and there was sufficient social and occupational data available for these miners Follow was based on continuing physical examinations annual company censuses questionnaires newspaper clippings employment offices vital statistics bureaus and the Veterans and Social Security Administrations . JACKSON 025544 104 Follow was 95 per cent complete as of the 1963 off date with all those of unknown status conservatively considered to be alive Informa- tion from death certificates autopsy reports and hospital and physician reports was used by two investigators to arrive at independent diagnoses for the 249 certified deaths among this group of miners Expected numbers of deaths for the group under study were computed from specific mortality rates for whites within the same states as the cases and total expected deaths were calculated for the period 1950-63 An SMR of nearly 400 22 observed and 5.7 expected - was determined for malignant neoplasms of the respiratory system The investigators capitalized on nearly 12,000 radon daughter the expected carcinogenic agent measurements made during the period under study by federal and state agencies When multiple measures for a given calendar year were available an average was used while measures - for adjoining years were used when data were not available These estimates were used to place each subject into one of six cumulative exposure categories and this was followed by an indirect age adjustment The analysis indicated that the standardized incidence per 10,000 increased from 3 to 116 as the cumulative exposure increased The reader should be cautious about selective factors in this study which may either invalidate or strengthen the conclusions drawn The 3415 miners included in this study are drawn from the 5270 miners the Colorado Plateau who volunteered for at least one physical and for whom sufficient data were available In another report on this group 122 to be reviewed next it was claimed that by 1960 90 per | cent of all miners in this area had presented themselves for a health examination although this estimate not been unequivocally verified JACKSON 025545 105 In the absence of verification it must be concluded that no well- defined cohort has been established which means that a precise denominator does not exist and SMR calculations are therefore questionable Also the authors state that members of the study group worked in 1200 different mines during the period under study Despite these objections the selective factors if any may well have resulted in an estimation of mortality from lung cancer which would serve to strengthen the findings reported | Using the same group described above Wagoner et al 122 stratified the 5370 eligible miners into one of five cohorts according to uranium exposure - white uranium miners and millers white millers white miners surface and underground among whom 317 deaths were observed during the study period Among white underground miners an SMR of approximately 400 for respiratory cancer was calculated In addition workers were further categorized according to duration of underground mining into one of two groups 1-4 years and 5 or more years Among the latter the respiratory cancer SMR was 1000 11 observed and 1.1 expected This same group of miners within a 15 county region of the Colorado plateau was used by Saccomanno 123 for a lung cancer study which utilized a case matching scheme Cases were divided into one of three groups 51 uranium miners who had mined for one week or longer and had lived in any of the counties 121 cases among all other male residents of the 15 counties and 37 transients who had either received medical care or who had died in the Colorado Plateau The 51 cases JACKSON 025546 106 among uranium miners were further stratified into one of three levels of cumulative radon daughter exposure The matched analysis indicated that miners got the disease at a younger age than miners that there was an excess of small cell tumor types oat round and polygonal cells - 57 per cent among miners against 16 per cent among the other cases and that the proportion of small cell malignancy types increased as cumulative exposure increased Archer 124 used the data from this study and others and had the cases reviewed by a qualified three man panel They found that epidermoid and small cell undifferentiated carcinomas as well as adenocarcinomas were all found in excess among uranium miners with small cell carcinomas the predominant type and the only one which was significantly elevated in all cumulative exposure categories They also concluded however that the relative frequency of histologic types is independent a conclusion which disagrees with the earlier report Archer et al 125 again used Colorado Plateau miners for an updated analysis in 1973 which includes 3366 white and 780 white miners of the original 5370 who are said to represent approximately 90 per cent of all miners in the area under study Miners included in this study had at least one month of underground employment prior to - 1964 with 99 per cent follow and mortality analyses for the period 1950-68 during which period there were 437 observed deaths against 277 expected The investigators then looked at respiratory cancer SMRs separately according to cumulative exposure 7 categories and years after start of underground mining 6 categories when the cases developed There was no excess in the lowest exposure category a significant excess in the next three exposure categories and an JACKSON 025547 107 increasing ratio of observed to expected in the two higher exposure categories These relationships prevailed across all years after categories A table analysis was used to quantitatively relate personmonths of exposure to observed lung cancer occurrence although the authors admit that there may be distortion in the same direction from two sources First miners worked throughout most of the study and therefore most of the latent period which means that if miner developed cancer while in the highest accumulated exposure category most of his person of experience would have been in the Tower categories Second most men in the lower exposure categories started mining more recently than those who contributed most to the higher categories leading to less elapsed latency time for months in the lower exposure groups Both distortions tend to maximize incidence in the higher accumulated exposure categories especially when assuming a long latency period in which case the table approach is not ideal Accordingly adjustments had to be made for latency period age and rate effect for which two approaches were used the calculation of observed and expected rates for year intervals after start of uranium mining which controls for latency period and use of a model which makes adjustments for time after the radon daughter exposure within each year o JACKSON 025548 SUMMARY OF OCCUPATIONAL STUDY DESIGNS It woulbde useful at this point to summarize the different methods and procedures utilized in the occupational studies reviewed in the previous chapter and to briefly comment on some of the benefits and drawbacks encountered in their use Study Designs and Measures of Effect A cohort design is generally most desirable for epidemiologic studies since this design allows for a direct determination of the ratio based on a true population at risk In occupational studies it is almost always necessary to utilize the retrospective cohort design because of the long latency periods of chronic diseases although at present there is much activity among industrial epidemiologists to initiate true prospective studies which would naturally involve - surveillance Notwithstanding the benefits of a cohort design a problem is often encountered in exposure allocation is how does one determine the exposure category to which a subject should be assigned In occupational studies it is generally rather straightforward since the choices are often dichotomons or exposed as for instance smoker or smoker The occupational milieu however often necessitates a judgemental determination of exposure status example the steelworker study utilized the reductionist approach For condensing 14,000 jobs to 75 subgroups and further into 10 basic groups JACKSON 025549 109 Workers however have a tendency to be very mobile within industries which means that they have often spent time in many different jobs with diverse exposures The steelworker study additionally generally required that a worker have a minimum of five years of exposure within an OT group before he was assigned to that exposure group The experiences of this group were then compared with the experiences of a comparison non- exposed group chosen from within the industry The trouble with this approach is that the individual assigned to the high risk OT has almost always worked in other OTs as well and is therefore not truly unexposed to the hazards of these other work areas He may even have worked for a number of years in the referent OT with which his experiences are being compared While this would tend to conservatize the measures of effect and might therefore be dismissed as only strengthening faith in our | conclusions we must be concerned with the possibility that this contamination may mask smaller differences which would otherwise be uncovered These methods then are very similar to the use within the rubber study of the most representative OT which classifies a worker into the OT in which he spent the greatest amount of time during his employ- ment within the industry Unfortunately this similarity results in similar drawbacks as well and should be interpreted cautiously ' Additionally when using the retrospective design one should be very careful to avoid differential outcome among steelworkers for instance it was found that the employers were unaware of 43 of deaths from malignancies and only 21 of heart and circulatory disease deaths Naturally these differences could cause havoc even the best designed studies JACKSON 025550 110 The use of the cohort design whether retrospective or prospective requires the very careful enumeration of the population PAR which is generally a highly expensive and difficult task As an alternative the control design can be used without PAR determination although the investigator must be very concerned with bias in case identification In a hospital based industrial exposure control study for instance such as the laryngeal cancer studies 116,117 among asbestos workers the researchers must take great care to make certain that cases of a certain type do not systematically seek care at a specialized hospital and thereby avoid the scrutiny of the investigators | When time money and extreme rarity of the disease are important considerations this design is often the one of choice We come now to the ambidirectional design which is the one utilized in the present investigation of stomach cancer but which was not used in any of the three industrial studies described in this chapter The ambidirectional design involves the identification enumeration and follow in this case accomplished retrospectively of a cohort of individuals for whom estimates of risk of various diseases are determined followed When an excess risk is identified it is by a control study to determine the specific exposures responsible for the excess Since this design is based only on a comparison of the experiences of the cases and referents and does not require detailed histories of all other unaffected members of the cohort it serves to economize exposure determinations although the necessity for exhaustive cohort identification remains The present study is not a truly ambidirectional ambidirectional study , however , but should more properly be described as a hybrid design . Had we utilized a true prospective cohort design the term ambidirectional would have been rigorously appropriate JACKSON 025551 111 Aside from these semantic considerations a number of additional comments should be made The study in this review most comparable to ours is the steelworker study because of the diversity of jobs within the industry In that study the ambidirectional approach has not yet been used because of the manner in which excess risks were initially identified While the rubber study first determined overall excess risks for specific diseases the steelworkers investigators determined these risks for specific OTS Once having identified a risk they then followed with a retrospective cohort design which compared the exposed members of the high risk OT with a carefully chosen unexposed another OT group while we must follow with a control study since the data necessary for the steelworkers approach has not been collected In all designs discussed a referent group must be chosen for . purposes of comparison Generally an attempt is made to chose a comparison group from within the industry under study to avoid any unforeseen differences or exposures and to avoid problems of the healthy worker type which will be momentarily discussed In most of the cohort studies cited the referent was another OT deemed to be unexposed for the hypothesis under consideration although this is by no means always the case For instance in one of the asbestos studies 114 a matched control group was chosen from a nearby mining town while the two laryngeal control studies 116 117 utilized hospital | controls The appropriateness of the referent group is not only important in analytic studies but may be of even greater significance in the initial descriptive studies on which the analytic investigations are based As mentioned earlier occupational studies generally begin with an assessment JACKSON 025552 112 of risk in the general industrial population or in some large groups within this population most often by use of the standardized mortality ratios SMRS or standardized proportionate mortality ratios SPMRS both of which require a referent or standard population Since the SMR is an indirect standardization specific if we are dealing with an age standardization disease rates for some standard are needed for the calculation of expected numbers of deaths Usually the most precise available estimates of these rates are for the general U. S. population and because of their accuracy and availability are often used The problems which the use of these industrial referents engender are well known and referred to as the healthy worker effect Briefly use of disease rates from these groups often leads to an underestimation of the SMR because an employed population by virtue of being employed is healthier than the general population which includes disabled and hospitalized people Therefore the observed disease frequency is lower and the expected frequency is higher in a relative sense and this often leads to low SMRs If the SMR is underestimated however important disease excesses may be missed entirely Ideally disease rates should be derived from other industrial populations which are free from the exposure under consideration Use of a referent of this nature would make the identification of a under specific disease excess within the unique population study more probable In the absencoef a disease rates for a suitable industrial referent population however general populations are used instead and are quite suitable as long as the investigator is aware of their limitations a JACKSON 025553 113 When enumeration of the PAR is not possible the SPMR is often used to provide some evidence of disease excess The known limitation of the SPMR however is that it indicates a relative and not absolute excess and can therefore by quite misleading Despite these drawbacks however there are instances where a specific proportional mortality ratio is so clearly elevated as in the early uranium miner studies 118 that concern about this limitation would not be appropriate Confounding and Effect Modification Epidemiologists are always concerned about the role of confounders in the estimate of effect especially when dealing with known confounders These known confounders are controlled either during subject selection or data analysis and the methods used for control of confounding in occupational studies do not differ at least in this literature review from methods generally used in other studies The confounders most commonly encountered are age race sex date of initial employment and total duration of employment which were all of importance in the present study as well The methods used for control of these confounders in the studies reviewed here include matching during subject selection and stratification in the analysis In addition there are available a number of multivariate statistical techniques which allow for the separation of the effects of confounders and true risk factors among which is Miettinen's confounder summarization score In the studies reviewed here matching was used in both cohort and control studies The intent of matching is to make the subjects as nearly alike as possible for extraneous factors so that any differences between subjects cases and controls or exposed and exposed will be JACKSON 025554 114 explained by the putative etiologic factors under investigation The major limitation of course is that once the investigator has matched determined on factor the effects of that factor can no longer be although by its nature of being a confounder the interest of the investigator is not focused on that factor anyway The other method used to control confounding was to stratify in the analysis without having matched during subject selection although it should be here pointed out that even when matching has been performed stratification in the analysis is still used When stratification without matching is used the idea to determine the effect of the variables under study within narrow ranges of the known confounders so that these confounders cannot have much of an impact on the measure of effect This method of stratification in the analysis was quite commonly used in the studies reviewed here In only one study where the combined effects of smoking and asbestos exposure in neoplasia were examined did the investigators look for interactive relationships or effect modification the latter being defined as the uniform association between exposure and disease across different categories of the modifier a term which is sometimes used interchangeably with interaction Selikoff and associates 126 report the experiences of 370 men who were alive as of 1963 94 of whom had died during the next four years and for whom there were available accurate smoking and asbestos exposure histories Of the 370 men under investigation 283 had smoked cigarettes regularly at some time Using a lifetable approach based on the U. S. general white male specific death rates for the calculation of expected numbers of deaths it was determined that only 47.5 deaths were expected compared JACKSON 025555 115 with the 94 that were observed Furthermore the excess deaths were due to bronchogenic carcinoma mesothelioma of the pleura and peritoneum asbestosis and cancer of the stomach colon and rectum The authors then calculated the expected number of deaths that would have occurred among this grouopf 370 men if their lung cancer mortality experience did not differ from other men taking both age and smoking habits into consideration The investigators found 24 deaths from bronchogenic carcinoma in this group while only 3.16 were expected a ratio of approximately 8 to , although they found no bronchogenic cancer deaths among men in this group who had never smoked regularly or who had only smoked cigars or pipes leading the authors to conclude that asbestos dust exposure without concomitant smoking does not increase the risk of bronchogenic carcinoma The authors further calculated that only 0.26 workers who had regularly smoked would have been expected to die of bronchogenic carcinoma during the period under study if they had neither smoked nor been occupationally exposed to asbestos dust while there were actually 24 deaths due to this cause This is the really ~ startling conclusion interactive effect of both smoking and asbestos exposure results in a nearly one hundred risk of dying from bronchogenic cancer compared with men who were not exposed to either This finding is an excellent illustration of the importance of searching for factors and when interaction when there are additional the appropriate data are available known risk Exposure and Latency There is an attempt in most occupational studies to estimate exposure qualitatively and quantitatively so that some determination ~ JACKSON 025556 116 of the relationship between exposure characteristics and disease occurrence is possible Generally an initial minimal exposure duration in years is necessary for inclusion in the exposed group For instance a minima year exposure was necessary in most of the steelworker analyses while a minimum of 15 years of employment was necessary for inclusion in one of the uranium studies 120 After the inclusion criteria had been set however there was then additional classification into categories of total duration in years or months This procedure was used in studies of all three occupational groups reported here Additionally in one of the steelworker studies involving cranemen 103 there was an attempt to qualitatively determine the type of exposure gaseous particulate or clean with which duration was correlated In two of the asbestos studies cited 112,115 there was some determination of dust levels In the first dustiness was assigned to one of three levels which were assumed to be interval data with level three having 3 times as much dust as level one In the second average dust exposures were assessed on a point scale based on both environmental measures and on interviews with knowledgeable employees Observed and expected mortality rates were then determined for each of the dustiness levels In addition to independent correlations between disease occurrence and duration or type of exposure these two characteristics are often combined sucahs in the asbestos study 112 with three dustiness levels where a weighted exposure average based on these levels and on total duration was used The most extensive determinationosf exposure were made for the uranium studies 127 where working levels WL and working level JACKSON 025557 117 months WLM were used The working level is a measure of concentration specifying an amount of radom daughters in a liter of air releasing a specified amount of potential alpha energy After the determination of the WL had been made the investigators then calculated the working level months These WLMS represent the cumulative product of the length of underground exposure in working months 170 hours and the concentration of radon daughters in working levels WL for specified mines and 7 calendar years | In most studies where latency period was examined it was calculated as elapsed time between first exposure and death Death was used as the endpoint because mortality data were generally used although these latency estimates can be rather misleading for diseases with long survival times or low fatality rates Since lung cancer however was the major type of malignancy under investigation in these studies these two limitations were probably not of major importance When latency was presented in a straightforward manner it was generally reported as a simple mean and range In one study 125 however | latency perioidn terms of years after the start of underground mining when the case developed was classified with cumulative exposures for a total of 42 categories in which were found an increasing ratio of observed to expected deaths in the two highest exposure categories a relationship which prevailed across all years after categories Summary A careful analysis of the studies cited above which are felt to be representative of generally good occupational studies reveals that there are certain basic characteristics of properly conducted investiga- tions in this field JACKSON 025558 118 First and most important is the careful definition and delineation of the cohort on which all subsequent determinations of excesses or deficits of disease occurrence will be based and from which generally are drawn all cases of a specific disease for retrospective such as case- control studies SMRs are consistently used for comparisons of observed and expected numbers of cases although the referent group varies from study to study and includes rates derived either from the U. S. general population from the geographic area nearest the site of investigation from within the study population or from comparable industrial groups with similar work characteristics but without the | exposure under investigation When possible precise types of exposure are determined e.g. particulate vs. gaseous preferably verified by environmental sampling Almost always disease rates are correlated with an estimate of duration of exposure an estimate which depends on availability of data and often takes the form of years or person of exposure to the agen under study Work experience is generally further categorized into different - levels of intensity of exposure often based on retrospective questionnaires Mean latency periods are also generally presented and they are usually based on elapsed time from first exposure When data are available occupational disease occurrences are compared across different geographic areas and across plants It should be remembered however that estimates of exposure are quite complicated when there exists more than one putative etiologic agent as is the case in the steel and rubber industries but which is apparently not a problem in the uranium miner and asbestos worker studies | Wagoner 121 has summarized the types of evidence needed to reasonably infer causality from an occupational study while at the JACKSON 025559 119 same time advocating the use of retrospective work histories and exposure data for the identification of occupational etiologic agents The steps are 1 demonstration of an excess risk 2 demonstration of a response relationship 3 persistence of both 1. and 2. when known confounders are eliminated 4 establishment of the consistency of the association 5 establishment of the specificity of the association 6 description of the histologically distinctive distribution of the disease especially among malignancies when possible according to site and type But even when all of these criteria have been satisfied associative conclusion should be made with great caution for as Donaldson 128 says epidemiologic studies can reveal the existence of a health hazard can define its distribution and dimensions in the population can identify associations which may be suggestive of etiology and can provide estimates of levels of risk under varying conditions Epidemiologic data will not - and should not be expected to - provide incontrovertible proof of a effect relationship This fact however makes neither the epidemiologic method nor the data derived therefrom any less scientific JACKSON 025560 MATERIALS AND METHODS Cohort Description All but three of the cases and all controls were selected from the ten year cohorts which were reconstructed for two companies plants in Akron Ohio These cohorts were retrospectively identified in very similar fashion at both companies and include all hourly factory active and retired employees who were 40-84 years of age as of 1 January 1964. The cohorts or populations at risk were followed for the ten year period 1964-73 and death certificates for almost all deaths among members of these cohorts were collected At Company A the cohort numbered 6678 | while These at Company B cohorts were the population at enumerated on the risk numbered 10,655 basis of a very thorough check of both management and union records and included a search of both payroll and pension records death claims severance awards and employment exits and transfers Deaths among cohort members were identified mainly from life insurance death claims but were checked both internally against company records and externally with the U. S. Internal Revenue Service Because of the company financed life insurance policy to which hourly employees are entitled after ten years of service follow is approximately 97 complete for each of the two companies In all during the ten year period under observation there were 1983 deaths at Company A and 2508 at Company B. JACKSON 025561 12 Case Selection Nearly all of the approximately 500 subjects in this study were chosen from the cohorts just described with the exception of three cases which occurred during 1974 or 1975. Before this present study was undertaken standardized mortality ratios based entirely on the cohorts described above and using the specific stomach cancer rates from the 1968 U. S. general population were determined after completion of nine years of the ten year period under study and we calculated an SMR of 187 at Company A and 123 at CompanyB for the full age range 40-84 the former including both white and black males while the latter examined the experience of white males only To correct for an causes SMR of under 100 among the group under study the stomach cancer SMR was divided by the causes SMR Table 1 to arrive at adjusted stomach cancer SMR which was 189 at Company A and 131 at Company B. Additionally since there was some objection to the use of 1968 U. S. national data for the calculation of SMRs during a relatively long ten year period stomach cancer SMRs were calculated separately for the five year periods 1964-68 and 1969-73 using national data from the midpoints of these quinquennial periods from which we derived SMR estimates which did not differ appreciably from these estimates already presented In this manner we established the existence of an excess of deaths in contrast with an race comparable group from gastric malignancies based on nosologized death certificate analysis within the rubber industry and we decided to conduct the present investigation to determine in which manner if any the work histories and other appropriate attributes of cases of stomach cancer differed from match d controls As mentioned above however case selection was expanded to include those which occurred in 1974 and 1975 as well as those which were identified during the interval 1964-73 JACKSON 025562 122 Despite the two year extension for case inclusion all controls were nonetheless limited to the initial ten year period under investigation reasoning that since the controls were to be matched on numerous factors with the cases the restriction of controls to the ten year interval would not in any way bias our conclusions so long as properly matched controls could be found for each case In addition death certificates for all '* controls through 1974 were checked to make certain that none of the controls ; had died from stomach cancer We examined the death certificates which arrived at OHSG and defined a case of stomach cancer as anyone whose death certificate anywhere included a diagnosis of stomach cancer either as an underlying or contributory cause of death 50 or when listed in the other significant conditions 6 section or as immediate cause 34 Initially 50 cases were identified among Company B employees and 54 among those who worked at Company A based on computer printouts requesting a listing of all individuals who had an ICDA code of 151.9 anywhere on the death certificate During a subsequent search of existing death certificate files for the two companies only 46 of the 50 92 of the CompanBy and 44 of the 54 81 Company A death certificates were located the others presumably being lost although our computerized records indicated the presence of stomach cancer somewhere on lost death certificates For the available death certificates Table 14 ~~ presents the place where the listing was found and Table 15 summarizes the terminology used to describe the cancer and the frequency of each type according to information abstracted from the death certificates An examination of these records indicated that one of the Company A cases was a carcinoma of duodeum and this was subsequently confirmed by tumor registry data while among Company B cases we found one listed as large bowel carcinoma with metastasis linitis plastica and another JACKSON 025563 123 TABLE 14 PLACE ON DEATH CERTIFICATE WHERE STOMACH CANCER WAS LISTED PLACE A Immediate Cause B Underlying Cause C Underlying Cause D Other Significant Conditions TOTAL NUMBER 34 31 19 6 90 PER CENT oe 38 34 21 7 100 TABLE 15 TERMINOLOGY USED TO DESCRIBE THE STOMACH CANCER PRIMARY Carcinoma TERMINOLOGY Adenocarcinoma Signet Cell Adenocarcinoma Esophageal & Gastric Carcinoma Mesenchymal Tumor Stomach Gastric Ulcer Cancer ^' Lymphosarcoma of Stomach Large Bowel Carcinoma with Metastases linitis plastica NUMBER 67 17 TOTAL 90 JACKSON 025564 124 as esophageal and gastric carcinoma , and one listed as gastric ulcer cancer with no confirmatory tumor registry information available for any of these The first of these four was excluded while the remaining three were included as unconfirmed cases It should be immediately noted that all cases were eliminated from the case pool when there was evidence from either death certificates or tumor registries that the primary was at a site other than the stomach Tumor Registry Confirmation As just described the selection of cases was based on a computerized listing of all cohort members who had an ICDA code 151.9 listed on their . death certificate in addition to the few cases identified from 1974-75 deaths This list was then checked against the records of the four largest hospitals in the Akron area Akron City Akron General St. Thomas and the Barberton hospitals The first three had hospital tumor registries while the last did not Each hospital searched for the records of all of the 105 cases on the list and this author personally abstracted the pertinent information at all hospitals except Akron City where the tumor registrar performed this task The information abstracted included the following when available date and method of diagnosis histological pathological description site of tumor within the stomach In all hospital records were found for 76 72 of the 105 subjects and either all or some of the desired information obtained although it should be noted that the acquisition of the record does not necessarily mean that a definitive diagnosis was made as for instance some records which indicated carcinoma NOS of the stomach based only on a clinical impression which does not differ much from an unvalidated opinion which might be found on some death certificate JACKSON 025565 125 At the Barberton Hospital which was the only one of the four without a hospital tumor registry this author examined the entire hospital record from which was abstracted the desired information At Akron General and St. Thomas the abstract prepared by the tumor registrar was used as a basis for information collection although these abstracts often did not include the site of the tumor within the stomach which was instead determinebdy an examination of surgical and radiological reports when possible This investigator was not allowed personal access to the records at Akron City where the data were compiled by the tumor registrar and although requested the site within the stomach is often missing because the registrar apparently limited herself to informa- tion on the registry abstract which generally does not request identification concerning the specific site within the organ with the primary When available however it was included Based on all the diagnostic information collected a four digit code was constructed and assigned to each case with the following information included method of diagnosis histologic type site within stomach and death certificate diagnosis Table 16 is included to indicate the range of values for each of the four variables and these codes will be used during analysis to differentiate between definite and possible cases the former including those confirmed anywhere in the hospital record Control Selection Control selection at both CompanyA and Company B was conceptually similar As discussed earlier in description of the design two sets of controls with two subjects in each set were selected for each case The first set was matched with the case on three factors age as determined by year of birth and age at death each within three years race and sex 1 JACKSON 025566 TABLE 16 CODING SCHEME FOR THE FOUR DIGIT PATHOLOGY CODE DIGIT I METHOD OF DIAGNOSIS CODE MEANING NO 1230 Histology Radiology 1230 Autopsy Clinical 1230 NSAurgery 9 UNK DIGIT II HISTOLOGY CODE MEANING Adenoca 12 Adenoca undiff or poorly diff 3458 Adenoca well diff Carcinoma 3458 Other NA 9 UNK DIGIT III DIGIT IV SITE WITHIN STOMACH DEATH CERTIFICATE DIAGNOSIS CODE MEANING NO CODE MEANING NO H Cardia & Fundus I Stomach Cancer upper third 2 Body mid third Pyloric region lower third Other Combination mentioned 8 NA - Other Primary } 9 UNK NA 9 UNK 126 JACKSON 025567 127 This group will be referred to as the ARS group The second group of two controls to be called the ARSD group was matched on duration of employment within three years in addition to the three factors on which the first group were matched It was also stipulated for the four controls of a given case that if the case was deceased the controls must have achieved at least the same age at death as the case with which they were matched in addition to the exclusion of deceased any control whose death certificate anywhere indicated a malignancy ICDA codes 140-239 . If the control was alive at the time of selection naturally this last stipulation was not applicable although the possibility exists that the control had in fact experienced a fatal malignancy of which this investigator would be unaware Two notes are appropriate at this time First the matching which involved calendar years year of birth age at death and employment duration in years was executed in disregard of months and days and may therefore result in an actual matching range greater than the stated one For instance the birth date of a case may be 1 Jan. 1900 while that of its matched control can be 31 Dec. 1903 resulting in an actual range of four rather than three years Since the maximum difference is therefore four years and the minimum zero the average difference was probably in the year range Second the duration used for matching is an estimate of that variable based on what is variously referred to as seniority or continuous service date These dates are determined by the individual companies based on a formula agreed upon by both union and management which subtracts from the actual starting date the amount of time spent away from the plant during unexcused or unrecognized absences These JACKSON 025568 128 seniority dates are therefore adjusted and not actual starting dates Because of the great amount of work which would have been necessary to calculate files for actual duration which is not already on existing computer the PARS under study this surrogate measure was used instead This drawback coupled with the use of calendar years discussed in the to preceeding paragraph can lead differences in actual duration which are considerably greater than the stated matching criterion of three years We will examine this during analysis of the actual work histories | to be described in the next chapter It should be immediately noted however that before the duration match was executed some preliminary analyses were conducted in one company to determine the difference between the two methods of estimating employment duration Table 17 presents the frequency distribution between duration determined by subtracting actual date in from date out and duration calculated by subtracting seniority date from date out This variable is called DURDIFF and as can be seen from an examination of the table the longer the duration of employment the greater the likelihood of a difference in estimated duration with an overall agree- ment rate of 61 The decreasing agreement between the two estimates as duration of employment increases may be explained in one of two ways first seniority dates were calculated by management in a manner which has become with time less stringent and more favorable to the employee and second the longer one has been employed the greater the opportunity for unexcused leaves of absence Not to be lost in this comparison of the two methods of determining duration of employment however is the fact that neither is an entirely accurate estimate of time duration The one which uses seniority data is JACKSON 025569 TABLE 17 DIFFERENCE IN DURATION DURDIFF WHEN CALCULATED ACCORDING TO TWO DIFFERENT METHODS" ACCORDING TO TOTAL EMPLOYMENT DURATION a PER CENT OF INSTANCES WHERE DURDIFF 77.9 81.5 79.6 39.5 27.6 . 61.0 TOTAL DURATION " 10 years 10-19 | 20-29 30-39 40 and over TOTAL ALL DURATIONS 129 aDURDIFF DUR 1 = DUR 2 i = DUR1 Date Out Date Out DUR2 - Actual Date In - Continuous Service Date Total Duration for this analysis is based on DUR 1 JACKSON 025570 130 dependent on the unique management agreements and methods of determination on which it is based while the estimate which makes use of actual starting date does not subtract from total duration the time the employewaes actually away from the job The analysis of the difference between the two methods described above was conducted on the total 1964 Company B cohort which includes over 13,000 individuals of all race combinations As noted the general agreement rate expressed as agreement within ranged from 1-40 years one year was 61 although the differences . The actual mechanics of control selection differed somewhat at the two companies At Company B lists of eligible controls for each subject were generated by computer from which the appropriate number of controls was randomly selected At Company A the entire procedure was executed manually which was a very time consuming process based on a liner printout which contains very basic information such as years of birth and hire social security number etc. of all individuals who were part of what is variously called the 20 or PAR sample which is an race | stratified random sample of the entire 10 year cohort on which the SMRS are based Control selection was limited to this PAR sample because the detailed work histories for the individuals included in this sample have already been entered on the computer presumably in a relatively final form although in actuality these work histories require so much editing and correcting that their utility in present conditioins far from ideal Since Company A deficiencies have already been broached control selection at this company will be described first As mentioned one- liner which included year of birth and death race and sex and employment duration based on seniority date for all members of the PAR sample was the basis of control selection First the year of birth of the case was noted and all subjects in the sample whose birth date was within three JACKSON 025571 131 years of that of the case and whose race and sex matched were blocked off and counted A table of random numbers was then used to choose randomly two of these eligible controls for the ARS group Within this same blocked off list those whose durations fell within three years of that of the case were then ticked off counted and once againa table of random numbers ordice ordice or a single die if the number of eligibles was small was used to randomly select two of these individuals as ARSD controls Before a subject was finally chosen as a control however two additional checks were made as stipulated in the chapter describing our design if the control was deceased he must have achieved at least the same age at death as the case with which he was being matched and he was not eligible if a malignancy of any kind was indicated anywhere on his death certificate This latter determination was made by an examination of the Company A death certificates which are here at OHSG and which are reasonably up to date through approximately 1975. Only when a control met all of these eligibility requirements was he accepted and quite a few were rejected because of an unacceptable age at or cause of death Additionally individuals were eligible as controls for only one case and once chosen could not be chosen again Some individuals were occasionally chosen twice but these errors were subsequently spotted and corrected Because of the numerous matching constraints however a full complement of controls could not be found for a few of the cases The procedure was conceptually similar at Company B although the initial stage of identification of all eligibles was computerized and the reservoir of eligible controls included the entire 10 year cohort and not just a sample of it For each case a printout was obtained which included all ARS eligible controls while those who were ARSD eligible as well were JACKSON 025572 132 so identified Included in this program were the same restrictions concerning age at and cause of death and those individuals who did not meet these criteria were eliminated a priori and did not even appear on the printout For each case the pages listing eligible controls were numbered and a random number table was used first to select page and then a subject this procedure being executed twice for the selection of the ARS controls For ARSD selection after a page was randomly chosen the number of starred entries on that page was tallied and either dice or a random number table was used for final control selection this procedure again being twice executed As can be seen therefore while the mechanics of control selection differed slightly at the two companies the matching criteria and restrictions were identical if one accepts as identical the use of either a stratified random sample of a PAR or the entire PAR as a control reservoir Because of these similarities in case and control selection as well as the availability of a referenced OT dictionary which transforms specific OTs into common or standard ones the possibility exists for combining strata a stratum is a case and its controls when appropriate during the analysis Additional Data Aquisition Much of the data necessary for this study were not available at OHSG and visits were made to corporate headquarters at both Company A and B in Akron As described above Company A . controls were members of the PAR sample whose detailed work histories had already been entered on the computer and while the cases were naturally not restricted to this sample almost all of them were on one file or another at OHSG In addition JACKSON 025573 133 the raw work histories which Company A sent to OHSG include place of birth so the combination of these serendipitous factors made the task at Company A headquarters rather simple Some minor problems were encountered however in trying to determine yearosf education from the work history folder face which listed only completed years of high school or college Since original applications were often unavailable this summary had to suffice although the interpretation of zero years of high school is moot since one cannot assume that the individual had completed junior high school or elementary or in fact that the individual had any education at all In addition the contents of some the very old personnel envelopes were destroyed after sometime and all that was kept was a copy of the folder face which lists only the first portion of the work history Also old pieces of work histories and indeed there are numerous pieces of work histories scattered throughout the folder were sometimes so aged that they could not be read or opened without tearing or breaking them Despite these minor problems however data collection proceeded quite smoothly and almost all necessary data were collected The problems at CompanyB corporate headquarters unfortunately were more numerous but most of them were more or less satisfactorily resolved before departure A brief overview of the record system may prove quite helpful in following this discussion and will therefore be presented first Personnel folders of employees who retired prior to 1962 are stored separately from those who retired after this date the former being stored in the basement at corporate headquarters and not subject to the same care in keep as the later retirees Specifically misfilinigs not uncommon and the aging of the folders themselves often lead to gaps in the work histories a problem which was encountered at Company A as well The JACKSON 025574 134' folders themselves are arranged in the following manner the work histories for the period 1911 when the plant was apparently opened to approximately 1930 were entered on sized cards kept in the folder jobs held between 1930-1962 were entered on the folder face and those held after 1962 are typed by some apparently computerized system onto special large sheets kept within the folders and the legibility of these different components differed substantially Least legible were those on the folder face since the dates indicating when an employee began a specific job were written at the extreme edge of the folder which was apparently subject to the greatest wear In cases where the dates could not be determined a careful estimate was made and when no additional evidence was available the midpoint between the preceding and succeeding dates was used The determination of education caused some interesting problems as well First there was often more than one employment application within the folder each of which requested information concerning education ina different manner In one the applicant was requested to indicate the number of completed years of education while the other just requested a tick next to either elementary junior or senior high school although it is not clear whether the tick indicated completion of that level or just some exposure to it And even in the latter case if the applicant indicated the completion of one year of high school is this nine or ten years of eduation While on this topic two additional observations will be related First there are very many Company B Akron employees from the neighboring state of West Virginia and in fact Akron is sometimes jocularly referred to as the capital of West Virginia Curiously almost every West Virginian indicated eight years of education almost as if they were instructed to do so Also the educational attainments of foreigners are JACKSON 025575 135 difficult to determine because over twenty foreign countries are represented and educational equivalents between these countries and the United States especially as long as 50-75 years ago are not readily available This conversion problem was encountered among the foreign- born at Company A as well The frequent existence of more than one application form caused problems in the determination of marital status as well since a number of years separated the two forms during which time marital status may have and often did change Since not all folders contained more than one application we used the status reported on the first application for the sake of consistency | A number of individuals had indicated in their personnel folders transfers to salaried positions sometimes permanently but generally for short periods of time These are generally all classified as simply transfer to salary although it should be noted that if the salaried position was that of a foreman or some similar job exposures actually continued during this period and we have no knowledge of the specific exposures involved Additionally there is no subtraction of time for isolated work stoppages and industry strikes which although probably involving relatively short time periods should perhaps be taken into account Economic Status SES Despite the fact that all subjects in this study are collar workers we were interested to know whether their economic status was similar especially since this factor has been associated with much of the 7 variation in stomach cancer incidence In 1949 Company A instituted a system of BATO JACKSON 025576 136 identifying each job with a digit code which along with the Department number uniquely determined each specific job within the plant Each of these jobs was then evaluated for twelve factors education experience skill initiative and ingenuity mental effort physical effort responsibility for material responsibility for equipment responsibility for safety of others responsibility for work of others work conditions and hazards An evaluation of each factor was made for every job and points were then assigned for each factor according to its importance the latter based on different weights for each factor In this manner a score was determined for each factor based on which a total score was then calculated by summation of the individual scores For the purposes of this study the scores for the first four factors and the total score were used to determine the skill level of the job which in turn was used as a surrogate measure of economic status on the assumption that the more highly skilled jobs generally pay the highest salaries Despite the imperfectness of this measure of SES we felt that along with education we would get a reasonable estimate of differences in the economic status of the subjects in this study It should be recalled however that this evaluation was initiated in 1949 so a substantial number of subjects had few if any evaluated jobs In addition some of the digit job codes were illegible on the work histories so even when an evaluation was available we occasionally could not use it and scores were not always found even when a digit code was present The existence of a similar system at Company is not well documented JACKSON 025577 137 Comments Concerning Some Variables on File Occupational Titles OTS are categories of jobs based on similar production processes and provide a mechanism for condensing the thousands of jobs within the industry into a manageable number for analytic purposes Company A uses a four letter OT code and has 110 OTs while Company B has 119 and uses a three digit code Because we wished to pool data from both companies for some analysewse needed a uniform OT system and since there is not a one correspondence between the OT systems of the two companies we had to use the Company A system which has fewer OTS Specific analyses were based either on OTS or OTGS Occupational Title Groups which number approximately twenty for standard use OTGs may however be created to reflect specific exposures a procedure which was used to determine odds ratios for carcinogen OTGs in this investigation Appendix I contains detailed descriptions of both the OTS and the standard OTGs An examination of the file format indicates two calculations involving duration- JOB DURATION and TOTAL DURATION The former is item specific indicates the number of days spent in the job which the item represents and is calculated by subtracting the date into that job from the date out on the basis of 30.42 days in a month and 365 days per year For instance if the dates in and out were 1/1/49 and 3/29/49 the JOB DURATION would be 89 days from which the reader should notice that weekends holidays and other work days are included This was done because we had no readily available way to determine how many real work days were included especially since week work was not uncommon TOTAL DURATION is simply the sum of JOB DURATIONS expressed in months based on 30.42 days per month in active OTs with durations from the following OTs excluded salary inactive retired sick lay off dead leave of absence miscellaneous absence exit JACKSON 025578 138 disability retirement and military duty One additional variable - AGE AT DEATH - requires comment here Examination of the file reveals a value for this variable for every subject in the study even when the date of death = 888888 not applicable or still alive For deceased subjects this variable has literal meaning | when the subject was still alive as of 1/1/76 this variable actually represents the subjects age on that date which is the termination date of the study Format An intermediate file including all available information for this study was created at OHSG This file was used for data analysis either directly or as the basis for creation of specific files for different types of analyses the latter including an SPSS file of fixed record Tength and a special file for the matched control analysis The format of the intermediate file is presented in Table 18 and a list of variables with appropriate explanations is shown in Table 19. Tables 20-22 show the coding Formats for marital status education and birthplace Analysis or SPSS _ SAS procedures were used for all statistical analyses reported in this study with the exception of the matched control procedure The latter computerized procedure was developed at the OHSG and a detailed description of it along with the algorithm on which it was based is presented in Appendix II For the packaged procedures the algorithms are available in the manuals from which they are derived The general format of the analysis is the presentation and discussion of the descriptive findings separately for case and control groups by company This is followed by a comparison of these findings which include JACKSON 025579 TABLE 18 FORMAT FOR THE INTERMEDIATE STOMACH CANCER FILE + US HA 2 Hf 99 99999 UCUDCD BLANK DATE DATEDATE BLANK BIRTHPLACE 9 9 99999 2 99 99 99 9 9 999999999999 8 999 99 99 99 999 9 9 9 9 916 99 99999 COMP A IN ITEM O.T. 1930 999999999399 9 9 9 9 939 32/334 32/3 4 36 37 38 40 41 42 44 45 47 48 L PATH AT AGEDEATHTOTALDURATIONEXPOSURAVY 999999 25 3024 3999999993 39999999999999999999999999999999999993999 9 9 9 39 9 93 44 50 51 29 31 32 33 34 36 37 40 41 42 44 45 45 47 50 53 54 55 56 57 58 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 76 139 JACKSON 025580 140 TABLE 19 1957 07 VARIABS VARIABS COMM NOMIERS , Variable Variable Col. Explanation Company number I tube~ 1 2-3 0 = Company A 1 == Company B Identifies a controls set case & 3 & 4 * ARBO : c umber Job Code SES education score SES experience score SES skill score 4 5-9 10-11 12-13 14-15 + 2 = AKS controls 5 digit code started in 1949 SES iniciation score 16-17 - Total SES score 18-20 SS 6 Sex , Race Educ M Stat Iten Company & OT Date In Date Out Dapt # Company B OT Jeb dur 21-29 30 : 31 32 33 34-35 36-39 40-45 46-51 52-55 56-58 59-62 male 2 = f male 1 = white 2 = black See coding instructions See coding instructions A history entry a letter code for specific item number for specific item number for specific item number a 4 digit code duration in days in specific item Birth Date Sen Date 63-68 69-64 Seniority or continuous service date 1st date ia 75-80 first date employed , Last date out Death date 81-86 87--92 last date employed 888886 if still alive UCD 54-97 ICDA code Birthplace Pathology Code 98-99 100-103 See coding format See coding controls format for cases 7777 for , Age at Death Total Duration Exposure Age 104-105 106-108 109-110 Age at deatli or age as of 01/01/76 In months for active OTS Age at Hire JACKSON 025581 TABLE 20 CODING FORMAT FOR MARITAL STATUS Company A 1 = single 2 = married 3 = separated or divorced 4 = widowed 9 = unknown Company B 1 = single 2 = married 3 = separated 4 = divorced 5 ==== widowed 9 == unknown 141 JACKSON 025582 TABLE 21 CODING FORMAT FOR EDUCATION Company A 1 = completed grade school 6 years Company B 1 = some or all grade school ( 6 years Company A and B 2 = some junior high school 7-8 years 3 - graduated junior high school 9 years 4 = some high school 10-11 years 5 = graduated high school 12 years 6 === some college or technical school 7 == graduated college 8 = not applicable no education 9 = unknown 142 JACKSON 025583 TABLE 22 CODING FORMAT FOR BIRTHPLACE 01 ALABAMA 37 PENNSYLVANIA 02 ARIZONA 38 RHODE ISLAND 03 ARKANSAS 04 CALIFORNIA 39 40 SOUTH CAROLINA SOUTH DAKOTA COLORADO 41 TENNESSEE 06 CONNECTICUT 42 TEXAS 07 DELAWARE 08 DISTRICT OF COLUMBIA 43 UTAH 44 VERMONTVERMONT 09 FLORIDA 45 VIRGINIA 10 GEORGIA 46 WASHINGTON 11 IDAHO 47 WEST VIRGINIA 12 ILLINOIS 48 WISCONSIN INDIANA 14 IOWA 49 WYOMING 50 ALASKA 15 KANSAS 51 HAWAII 16 KENTUCKY 52 ARMENIA 17 LOUISIANA 53 GERMANY 18 MAINE 19 MARYLAND| 20 MASSACHUSETTS 54 SYRIA ENGLAND CZECLOSLAVAKIA 21 MICHIGAN AUSTRIA 22 MINNESOTA ITALY 23 MISSISSIPPI 24 MISSOURI 25 MONTANA HUNGARY HUNGARY RUSSIA 26 NEBRASKA 27 NEVADA 28 NEW HAMPSHIRE TURKEY YUGOSLAVIA RUMANIA 29 NEW JERSEY 30 NEW MEXICO 31 NEW YORK 32 NORTH CAROLINA 33 NORTH DAKOTA 34 OHIO 35 OKLAHOMA 36 OREGON POLAND W. UKRAINA SCOTLAND LITHNANIA IRELAND SWEDEN GREECE FRANCE USA NOS 143 JACKSON 025584 1.44 frequency distributions for all variables and descriptive statistics including mean median standard deviation and standard error of the mean for continuous variates The findings for all study factors other than OTS and OTGS are presented in the first analysis chapter followed by the presentation of findings from the detailed work histories in the | second The basic procedure used for analysis was an examination of effect measures for various factors and the presentation of the differences in the distribution of variables between cases and controls The former was accomplished mainly through extensive use of the Haenszel Odds Ratio OR for matched studies with a variable number of controls per case and is the control equivalent of the relative risk estimate derived from cohort studies As such it can be considered a measure of effect althougwhe have not strictly speaking directly measured this effect The presentation of the odds ratio and its associated confidence interval is descriptive and for this investigation the effect measure was not considered statistically significantly elevated if the 90 interval included one which is the null value for an odds ratio or relative risk The confidence interval itself was based because it was calculated using the formula which is derived from the data or test of hypothesis being conducted The major procedure for the second type of analysis where distributions and summary statistics of variates are considered was the Paired Test This procedure was the one of choice and was used whenever possible because it makes full use of continuous data and because it is a valid . statistically procedure for the analysis of data generated from a matched design The former reason is quite important in light of the fact that utilization of the program available at the OHSG for the calculation JACKSON 025585 145 of ORs and their associated tests of significance forces the investigator not only to categorize but to dichotomize the exposure or the putatively causal factor a procedure which leads to loss of information and which is anathema to epidemiologic investigations While the Paired Test does not utilize a direct equivalent of an OR it does not at the same time lead to inefficiency and does allow for a statistical test of the significance of the mean differences of the variable between the groups being compared this latter being an indirect measure of effect It should be added here that the Test was conducted in a manner which differed somewhat from the usual procedure because the difference between a case and its controls was based on an average value for the controls Since all cases did not necessarily have a full complement of controls the average is not based on a constant number of controls which means that the assumption of normality is not being met precisely but this slight departure from normality was not considered very damaging because of the robustness of the Paired Test procedure The major measures of effect then are odds ratios and comparisons of distributions and summary statistics such as means When using these procedures control for confounders was accomplished by stratifying on : these factors in the analysis and examining the odds ratios or values for the Tests at different levels of the confounder bearing in mind that control for some potential confounders age race sex duration was accomplished in the design by matching on these factors For instance an OR for a specific exposure was calculated separately for native and foreign subjects and if these ORs were pragmatically similar an overall nativity OR was then presented It was felt that a test for homogeneity of odds ratios across strata would not be very helpful since we were more interested in practical rather than statistical JACKSON 025586 146 significance when considering the pooling of strata for the presentation specific of findings When rates differ considerably the _ reader's attention is directed toward this difference and the measures of effect are presented separately We were naturally concerned with interactions between factors and their possible effects and while statistical tests for interactions were generally not used decisions were based on an assessment of their practical effects after examining different combinations of the interacting variables For instance after determining that an OR differed neither between sexes nor between nativity groups the ORs were examined for combination of these variables e.g. female native female foreignborn male native etc. and decisions concerning data pooling were based on these examinations Two final comments regarding statistical tests are in order here First tailed probability tests were most often used because of the generating nature of this investigation resulting in our inability to consistently predict the nature and direction of the differences Second 90 and not 95 per cent confidence intervals were calculated for the odds ratios so occasionally an interval whose lower | bound is greater than unity is not starred as significantly elevated since the chi statistic is used for the purpose of establishing statistical significance JACKSON 025587 ANALYSIS I DEMOGRAPHIC AND OTHER FACTORS This chapter will present the general descriptive findings along with the analytic results which do not deal directly with the detailed work histories on which both exposures and OTS Occupational Titles are based These latter will be exhaustively examined in the following ~ chapter Matched Variables Subjects were matched on factory A or B age year of birth and age at death both within three race sex and for one control group ARSD duration of employment An edit program run on the study file indicated that the matching constraints had been executed properly during the subject selection procedures resulting in similar distri- butions for these factors among both cases and controls It is of interest nonetheless to present these distributions in describing the general attributes of the subjects included in this study In total the study included 495 subjects- 52 cases and 197 controls in Company A and 50 cases and 196 controls in Company B. While the original design called for four controls for each case suitable comparands could not be found in 15 instances as is obvious from the totals presented above Despite this inability cases were not excluded unless it was impossible to find at least one control for each of the two control groups Overall 46 subjects or approximately 10 per cent were females including 10 cases and 36 controls From this can be seen that it was JACKSON 025588 148 relatively more difficult to find controls for females since they comprise 10 per cent of the study population but they account for 25 per cent of the unsuccessful attempts at matching which is probably due to the rather small pool of women employed in the industry A slightly smaller proportion of subjects 41 or 8 per cent were black and this group was comprised of 9 cases and 32 controls All blacks in this study were males and a disproportionate number of these subjects 32 had been employed by Company A. More than half 8 of 15 of the missing controls therefore were not found for cases who were either female all of whom were white or black males Table 23 summarizes the race and sex distribution of all subjects Overall 77 percent of cases and 83 per cent of all controls were married at the time they were first employed and the specific rates are quite to similar these as an examination of Tables 24-26 indicates It should also be noted that in both companies the proportion of cases who were separated or divorced and the proportion widowed was approximately twice that of controls although the numbers in these categories are quite small At the time of death from available death certificates 68 per cent of Company A cases and 75 of those in Company B were still married while 27 and 22 per cent respectively were either widowed or divorced Education There were no striking overall or specific differences in educational attainments of cases and controls as an examination of JACKSON 025589 149 Tables 27-29 will reveal with about 75 per cent of both groups having completed only junior high school or less In general Company B subjects completed more years of schooling than those in Company A especially as regards the proportion who completed their high school education This may be partly due to the fact that 7 of the 9 black cases and 25 of the 32 controls in this race group were employed at Company A and only one member of this group had graduated junior high school with the great majority of the remaining ones having a grade school education or less Cases in both companies tended to have slightly fewer years of schooling than controls with most subjects having formal education less than high school It should be remembered however that many of the subjects in this study were of school age early in the century when high school completion was not common a fact which along with the collar status of employees doubtless explains the reported levels of formal education received It should also be mentioned that a fair proportion of subjects had been born in foreign countries most of whom were born in Europe where one year of education may be more than the equivalent of one year of American education It should finally be noted that females as expected had received less education than males but since the number of females was approximately equal in both companies this had no effect on the differences in specific educational attainments We also conducted a further analysis of Economic Status SES by using Job Codes within Company A with which were associated evaluations on 12 factors including education skill initiative and experience These codes were introduced in 1949 so a relatively small proportion of jobs within our study were eligible for inclusion in this analysis After JACKSON 025590 150 extensive efforts in creating the file for this analysis however we observed a score for only one of the 52 cases within this company while over half the controls had scores Because of the dearth of scores among cases the analysis was impossible Life Status and Age at Death Stomach cancer case identification for this study was based on mortality data so all cases are deceased The age match however was based on two factors year of birth +3 and age at death for which the control must have been at least as old as the case +3 if the control was deceased According to this constraint however the controls could have still been alive at the time they were selected and many in fact were Table 30 presents the distributions of age at death for the cases and decedent controls and of age as of 1 January 1976 for living controls as well as the number and proportion of controls in the latter category As is seen 75 per cent of the controls were still alive on the closing date of the study and many in this group have already attained very advanced ages In addition a greater proportion of the deceased controls attained older ages than did the cases These findings as indicated are doubtless explained by the matching constraints The ages of subjects at death were further analyzed according to pathology and the mean among confirmed cases 70.5 was virtually equal to the mean age at death among the unconfirmed cases 71.5 In addition the cumulative proportions of cases dying by the age of 65 were nearly equal in the pathology groups 22 vs. 21 per cent as were the cumulative proportions dying by the age of 80 91 vs. 84 per cent Overall the mean age at deatohf Company A cases was 70.4 while among those at Company B the average age at death was 71.4 JACKSON 025591 151 Place of Birth Ascertainment of the birthplace of subjects in this study was exceptionally good with only one unknown among the 495 subjects Overall 22 per cent of all cases and 17 per cent of controls were foreign born although there were differences in these proportions between the two companies In Company A 25 per cent of the cases and 23 per cent of all controls were foreign born while the respective proportions in Company B were 20 and 12 per cent a finding which may indicate different hiring policies at Company A. For both companies then the proportion of foreign born among cases was greater than among controls with this trend holding for three of the four control groups the only exception being the ARSD control group at Company A where 32 per cent of the controls in this group were not born within the United States It should also be noted here that three of the ten female cases and seven of the thirty six female controls were foreign born and these proportions are quite similar to those among males In addition as expected all black subjects were native born- Nativity is a factor of much importance in this study sinciet may | by itself account for the excess of mortality from gastric cancer among employees of the rubber industry further analysis of this variable is therefore appropriate The list of foreign countries in which subjects were born Table 31 ) which numbered 21 was divided into tertiles according to mortality from stomach cancer using Doll's ( 3 ) male adjusted international mortality rates for this disease High risk countries had death rates of 75 per hundred thousand and above intermediate risk countries had rates between 50-74 and those classified as low risk countries had death rates below 50. Before examining the JACKSON 025592 152 results however numerous limitations should be stated First the mortality rates used for creation of the risk categories are based on standardized rates for males between the ages of 45 and 64 years In our study however only 22 of the 102 cases 22 per cent died at an age below 65 leaving open the possibility that the ranking procedure might have differed from the one used if the standardized mortality experience for all and not just the younger age groups had been used as the basis of the categorization Second assumptions concerning two of the countries on the list were made people born in Germany were assumed to have been born in what is presently West Germany while those subjects born in Hungary were assumed to have mortality rates somewhere between those of what is presently Austria and Hungary An examination of the distributions presented in Table 32 reveals very little difference in nativity risk between cases and controls with 48 per cent of cases and 46 per cent of controls having been born in countries categorized as being at high risk of mortality from stomach cancer Similarly the proportions born in countries with moderate and low risks are also essentially equal Hasty conclusions should not be drawn from these data however since the factor of most importance is that a greater proportion of cases were born in foreign countries and all of these countries have higher mortality rates than those of the : United States Because of the limitation concernitnhge age range of mortality | rates on which the risk stratification was based use was made of World Health Organization 129 ) specific data for the available 13 of the 21 countries in which subjects were born Table 33 presents mortality rates for age groups 65-74 and 75 and above The data for I each of these age groups was used independently to generate risk JACKSON 025593 153 rankings and a comparison of these separate distributions is presented When using the younger age range as the basis for the rankings the categories are almost exactly the same as those based on Doll's rates while there are appreciable differences in the ranking of countries based on the older age group In addition the proportion of cases and controls in each of the risk categories in the categorization based on mortality among the younger males are essentially equal and very similar to those which use Doll's data although there is some divergence of proportions in the ranking based on mortality among those aged 75 and | over When these data were analyzed separately for confirmed 64 cases and unconfirmed 38 cases the results indicated that 17 27 per cent of the confirmed cases and 6 16 per cent of the unconfirmed cases had been born in foreign countries Using the risk categorization based on Doll's data we get the results presented in Tabl3e4 , from which we can see that among confirmed cases and their matched controls a greater proportion of cases 9 or 53 per cent were born in high risk countries than their matched controls 17 or 40 per cent in addition to which a . greater proportion of confirmed cases 53 per cent compared with those who were unconfirmed 33 per cent were born in countries at high risk of mortality from stomach cancer When confirmed cases and their matched controls are further analyzed however we see that the cumulative proportion through the moderate category are essentially equal 88 vs. 85 per cent while among the unconfirmed cases and their matched controls the cumulative proportion in these two categories is higher for the controls 66 vs. 85 per cent The birthplaces of native American subjects were further examined to see if cases were more likely to have been born in states with JACKSON 025594 ; 154 higher death rates As expected the greatest proportion of subjects was born in Ohio This both cases and controls proportion is 22 separately The per two cent and is the same states in which the for next greatest proportion of subjects was born are West Virginia and Pennsylvania in which a combined proportion of 31 per cent of the cases and 26 per cent of the controls were born In fact approximately half of all subjects had been born in one of these three states All states in the continental United States were then categorized into one of three groups according to death rates among white males ( 5 ) and the results of this stratification are presented in Table 35 As can be seena slightly higher proportion of both confirmed and unconfirmed cases had been born in high risk states although the differences between these proportions and the differences in death. rates between the states are not striking and there does not appear to be much difference in the distributions according to pathology status as well It should be mentioned here that one might object to all the categorizations of risk presented here because the mortality rates on which they are based are all rather recent while the subjects had left these areas for the most part long ago This objection however is not disturbing because the relative rankings and not the absolute values are the major concern and these rankings seem to be rather | ~ consistent temporally One final analysis was performed because of Mancuso's ( 130 finding that there is an excess of deaths from stomach cancer among the southern who have migrated to Ohio Among subjects from Company A 17 of the 52 cases 33 per cent and 85 controls 43 per cent had been JACKSON 025595 155 born in the South South Atlantic East South Central West South Central while the comparable figures for Company B were 34 and 38 per cent respectively In summary then the only important differences in nativity between cases and controls is in the overall proportion born in foreign countries although even this difference is not great Among these subjects however approximately equal proportions were born in countries with high intermediate and low mortality rates from stomach cancer while among native born subjects the same conclusions can be drawn regarding states Underlying Causes of Death The inclusion criteria for cases in this study was the listing of stomach cancer anywhere on the death certificate As expected however it was listed most frequently 89 of 102 or 87 per cent as the underlying cause of death as reference to Tables 36-37 indicates with the second most frequently mentioned underlying cause of death being ischemic heart disease ICDA 410-414 which appeared on five per cent of the death certificates Controls were eligible for inclusion only if there was no mention of a malignancy anywhere on their death certificates if deceased Examination of Table 37 reveals that of the 96 decedent controls 56 58 per cent had ischemic heart disease mentioned as the underlying cause of death followed by cerebrovascular diseases ICDA 430-438 diseases of the arteries ICDA 440-448 and emphysema ICDA 492 that 76 of the decedent controls 79 per cent died of circulatory diseases The disproportionate number dying from ischemic heart disease is not surprising in light of the control selection restriction regarding malignancies JACKSON 025596 156 Pathology Data As described earlier the four major hospitals in the Akron area to which cases of stomach cancer were likely to have gone for treatment were searched for records of the cases included in this study and we were able to gather some confirmatory evidence for the disease as stated on the death certificate for 64 of the 102 63 per cent cases included in this study with confirmation based on evidence from either autopsy histology radiology surgery or clinical impressions In addition to this information data on type of cancer and site of the disease within the stomach were collected and these findings are | Summarized in Table 38. In examining the section describing the methods of diagnosis it should be kept in mind that the order of precedence for the listing presented is autopsy histology surgery radiology and clinical impressions An examination of this table indicates that 56 of the 64 88 per cent confirmed cases were verified on the basis of either pathology or autopsy reports the two methods generally thought to be the most reliable method of diagnosis In addition 55 of the 69 cases 86 per cent for whom the specific diagnosis was known were classified as having adenocarcinoma of some form including well or moderately differentiated and poorly or undifferentiated adenocarcinoma a finding which agrees closely with other published reports including Ming's ( 75 ) findings cited earlier The finding that 18 of the 35 cases 51 per cent for whom the specific site within the stomach was established had the lower third or pyloric region indicated was also in general agreement with Stout's 76 ) reported findings When these factors were examined separately by company there was no overall difference in the proportion JACKSON 025597 157 of verified cases but among subjects for whom pathology data were available a greater proportion of Company A Subject1s2 of 34 or 35 | per cent had the cancer in the pyloric region of the stomach than did those at Company B 6 of 31 or 19 per cent Calendar Year of Hire Calendar year of hire is an important factor because of the possibility that gastric carcinogens may have been either introduced or eliminated in different calendar years Any significant differences between cases and controls therefore in year of hire would be of | interest and would require further examination Overall approximately half of cases and controls had entered the industry by 1929. This was followed by a relatively small proportion entering during the next decade and a large number entering during the period 1940-1949 almost all of whom had begun working in the industry during the war years probably immediately after discharge from the service While the general pattern of entry is similar in both companies there are differences worth noting At CompanyA the largest proportion of both cases and controls 40 per cent entered during the 1920 decade followed by the 1940 and 1910 decades At Company B on the other hand entry was much more common during the years 1910-1919 approximately 40 per cent and this was followed by entry into the industry during the 1920 and 1940 decades It should be noted here however that while there were major company differences the proportions of cases and controls beginning their employment within the industry within each JACKSON 025598 company separately were quite similar as an examination of Tables 158 39-41 will reveal except for entry into Company A during the first decade It should also be noted that none of the female cases and no black subjects both cases and controls had entered the industry during the first decade 1910-1919 of plant operations at both companies To further analyze for differences in this important factor Paired tests were performed on the groups to be described with the slight modification of using an average for the controls in any given stratum a case and its matched controls and a tailed test of significance was used since we could not always predict the direction of the expected difference Results of this analysis are presented in Table 42 from which it can be seen that only in the comparison of cases with ARSD controls at Company A were the differences in year of hire _ statistically significant and this difference isprobably reflected in the different entry rates for the first decade of plant existence Further analysis of the sex specific differences and their probabilities revealed that statistical significance for this difference was maintained among confirmed white male cases at Company A only p = 0.022 and not for white females and black males although the numbers in the latter two categories are small While the difference in entry year between white male cases and controls did achieve statistical significance however the practical importance of this difference 3.0 years is questionable although it should be added that this finding was initially interesting because it occurred in the comparison of cases with matched controls the D of course signifying that these controls were matched on total duration of employment as well as on the other three factors This interest diminishes however because of the ; JACKSON 025599 159 statistically significant differences for the comparisons with the ARS group and with both groups in Company B and because while the group was matched on duration the estimate of duration was indirect and was based on a surogate measure as described in Materials and Methods Further analysis according to pathology confirmation status did not alter any of the findings described and the distributions are presented in Tables 43-44 Generally there were significant differences between cases and controls within both the confirmed and unconfirmed groups with cases tending to have started slightly earlier than controls A comparison across these groups revealed that unconfirmed cases had started their employment within the industry at a slightly earlier calendar period than those cases whose illness was confirmed After these analyses had been completed we decided to look a bit more closely at the group in Company A comprised of 11 cases and 21 controls which had entered the industry during the first decade 1910-1915 of existence of the plant in Akron because there had been a disproportionate number of cases in this group as reference to Table 39 will confirm Because the only significant differences within Company A occurred in the comparison of subjects in this company with ARSD controls the following description will be limited specifically to this group All subjects in this group were white males Five of the 11 cases 45 and 8 of the 20 ARSD controls 40 were foreign while only 8 of the remaining 41 cases 20 and 23 of the 76 controls 30 % who entered the plant after the first decade had been born in countries other than the U.S. It is therefore apparent that all subjects including both cases and controls who entered the plant during the first decade were more likely to have been foreign than those who entered after the first decade BATO JACKSON 025600 160 The analysis of other important factors reveals striking differences both between cases and their matched controls who entered during the years 1910-1919 and between subjects who entered during this period compared with those entering later with a summary of these findings appearing in Table 45. As can be seen first decoa f ed nte ry cases worked approximately 70 months longer than their ARSD matched controls 385 vs. 317in active OTs and began at an age approximately 9 years younger 25 vs. 34 In addition to these findings we also see that these cases worked longer and started younger than the remaining cases as well While the age at death of these cases was about 8 years older than the remaining cases we must remember that the design called for inclusion of cases who had died only during the decade 1964-1973 Since these cases started earliest they would be expected to have lived longest Age at death of controls is not presented because most were still alive and because the constraint was that they must have lived at least as long or longer than the case making a comparison meaningless Further examination of this group revealed one additional interesting finding The overall significant differences between cases and ARSD controls in age at hire and year of hire are accounted for entirely by this group which entered during the first decade When this is further examined by pathology confirmation status these significant ' differences persist only among the 6 confirmed cases which means that the overall significant differences described earlier are explained entirely by this one small group of cases and their matched controls These results are summarized in Table 46 . JACKSON 025601 161 Total Duration of Employment As just reiterated the matching procedure for total duration of employment was based on the difference between the last year employed and the seniority year which is a reasonable but not an exact measure of duration For this analysis however total duration was based on time spent within active OTs which excludes sick leaves leaves of absence military duty etc. Tables 47-51 present the distributions for these variables in months from which it can be seen that the overall means 358 vs. 353 and the medians 386 vs. 369 are remarkably similar as are the general distributions of the categorized values for this factor Once again however there are some differences between companies with subjects from Company B having somewhat longer durations than those from A although the company differences between cases and controls like the overall differences are rather small and inconsequential The results of the tests are presented in Table 52 $ and an examination of these findings again indicates that the differences in total duration of employment in active OTs is neither statistically nor practically significant In fact the greatest discrepancies albeit small were found in the comparison of cases with ARSD controls again noting that this group was matched on duration of employment Once again the explanation probably lies in the fact that the matching procedure was based on an indirect determination of the value for this variable When these data were further examined according to pathology status we found that confirmed cases had worked slightly longer than unconfirmed cases but this difference did not appear to be of practical significance JACKSON 025602 162 while the differences within groups were neither of practical not statistical significance Despite the fact that one control group ARSD was matched on duration of employment while the other ARS was not the duration within active OTs did not differ much between cases and both control groups separately and combined nor did it differ much between control groups themselves To test this paired tests were performed on the difference between the mean total durations in active OTS of the ARS vs. ARSD control groups These differences were minimal and not statistically significantly different Accordingly in subsequent analyses the distinction between the two control groups will be ignored Age at Hire Age at hire is a factor of particular importance because of the generally held belief that the younger the age at which exposure to a carcinogen begins the greater the risk of developing cancer Tables 53-57 display the proportions of subjects in different categories of this variable and we once again observe remarkable similarities in the means and medians of this variable in both overall and company specific findings with slight differences between companies In general subjects began working in the industry around the age of 30 with Company B subjects beginning work at a slightly younger age than those at Company . A the difference being approximately three years A closer examination of the distributions however reveals that the cumulative frequencies for Age at Hire before the age of 29 are greater for cases in both companies and the proportions of cases hired at the youngest ages are JACKSON 025603 163 also greater than the corresponding proportions for controls These apparent differences mostly disappear however when results of the Paired test presented in Table 58 ' a procedure which maintains the integrity of the matched strata are examined . These results again indicate no statistically significant differences between cases and controls except for the comparison of cases with ARSD controls at Company A where the difference approaches statistical significance p = 0.02 but probably not practical significance Further analysis as with the previously discussed factor again indicates that a statistically significant difference in this group was found only among white males p = 0.027 and not among black males or white females It is worth noting however that while the differences between cases and controls were indeed slight cases had been consistently hired at younger ages than controls As with the variables described previously separate analyses for confirmed and unconfirmed cases and their matched controls did not reveal any appreciable differences and whatever discrepancies did exist were not statistically significant JACKSON 025604 TABLE 23 RACE AND SEX DISTRIBUTIOONF SUBJECTS All Cases Males Females Total Black Total 83 9 92 10 1 10 - 93 9 102 All Controls Males White 325 Females 72 Total 361 Black Total 32 357 ~ 36 32 393 164 JACKSON 025605 165 TABLE 24 MARITAL STATUS OF SUBJECTS AT EMPLOYMENT 1 COMPANY A CASES VS. CONTROLS Marital Status : Single Married Div Widowed Totals Cases Frequency Per Cent 8 . 15.4 37 71.2 3 5.8 4 _ 7.7 52 100.0 Controls Frequency Per Cent 20 10.2 167 84.8 6 3.0 _ 4 197 2.0 100.0 JACKSON 025606 | i 166 TABLE 25 MARITAL STATUS OF SUBJECTS AT EMPLOYMENT 2 COMPANY B CASES VS. CONTROLS Cases wee ee en ee ee ee ne er Controls Single Married Div Widowed Unknown Totals | 0. 41 3 4 2 50 0.0 82.0 6.0 8.0 4.0 100.0 14 158 9 8 7 196 7.1 80.6 4.6 4.1 3.6 100.0 JACKSON 025607 167 TABLE 26 MARITAL STATUS OF SUBJECTS AT EMPLOYMENT 3. ALL CASES VS. ALL CONTROLS Marital Status - Single Married Div Widowed Unknown Totals Cases Frequency Per Cent 37 49 36.3 48.0 6 5.9 8 7.8 2.0 102 100.0 Controls Frequency Per Cent 181 - 178 15 12 7 393 46.1 45.3 3.8 3.0 1.8 100.0 JACKSON 025608 168 ee em ee re een ee TABLE 27 EDUCATIONAL ATTAINMENTS ATTAINMENTS OF SUBJECTS 1. COMPANY A CASES VS. CONTROLS Cases Frequency Per Cent Controls Frequency Per Cent Grad Crade Schl Some Jr. High Grad Jr. High Some High Schl Grad High Schl Unknown Totals _ 26 50.0 10 19.2 13.5 1.9 117 -36 18 59.4 18.3 9.1 2.5 0.5 10.2 100.0 JACKSON 025609 TABLE 28 EDUCATIONAL ATTAINMENTS OF SUBJECTS . COMPANY B CASES VS. CONTROLS 169 Education Grad Crade Schl Some Jr. High Grad Jr. High Some High Schl Grad High Schl Some Coll Tech No Education Totals Frequency Per Cent 6 12.0 24 48.0 3 6.0 5 10.0 3 6.0 1 2.0 2 4.0 Frequency = Per Cent 26 13.3 83 42.3 17 8.7 26 13.3 22 11.2 1 0.5 1 0.5 50 100.0 196 100.0 JACKSON 025610 170 TABLE 29 , EDUCATIONAL ATTAINMENTS OF SUBJECTS 3 ALL CASES VS. ALL CONTROLS pee ee ee Education Education Grad Grade Schl Controls eres ce ects Cases ce - - Frequency Per Cent Frequency Per Cent 32 31.4 143 36.4 Some Jr. High Grad Jr. High Some High Schl Crad High Schl Some Tech No Education 34 10 6 3 1 3 33.3 9.8 5.9 2.9 1.0 3.0 119 35 31 23 1 1 30.3 8.9 7.9 5.9 0.3 0.3 Unknown Totals 13 102 12.7 100.0 40 393 10.210.2 100.0 JACKSON 025611 ** . a } TABLE 30 LIFE STATUS AND AGE AT DEATH OF SUBJECTS 171 Alive Dead Total 295 98 393 75 25 100 - 5 Cases Age at Death Controls Age atDeath Controls Age on 1 Jan Age Per Cent # Per Cent } Per Cent - - - - . 40-49 1 1.0 - 50-59 13 12.7 4 4.1 12 4.1 ; " 60-64 00 7.8 2 2.0 15 5.1 65-69 18 17.6 6 6.1 35 11.9 70-74 24 23.5 17 17.3 54 18.3 75-79 1.80-84 23 13 22.5 12.7 3.1 31.6 29 29.6 56 19.0 82 27.8 85+ Totals 2 102 2.0 100 9 19.2 98 100 41 295 13.9 100 4 JACKSON 025612 172 TABLE 31 BIRTHLACH BIRTHLACH BIRTHLACH ALL CASES VS. ALL CONTROLS CONTROLS Birthplace U.S. Foreign Armenia Austria Austro CzechoslovakiCzecahoslovakia England Walea France Germany Greece Hungary Ireland Italy Lithuania Poland Ruaania Russia Scotland Sweden Syria Turkey W. Ukrainia Yugoslavia Controls Frequency Per Cent 82.7 ~ 37.3 17 JACKSON 025613 173 Country Russia Poland Czuchoulvakia Hungary Runanda Ukrainia Austria Hungary Totels Ita'y Germany Lithuania Scotland England Totals Yugoslavia Ireland France Armenia Sweden Creece Totals TABLE 32 CATEGORIES FOR SUBSYS SUBSYS BRU BRU IN IN STOMACH } FRON STOMACH CANCER HIGH Risk Countries 89R 75 SDR Cases Cases Number Per Cent Controls 128 90 $ 86 84 77 75 75-86 22-2 22-2 22-2 22-2 | 1 3 for 11 8.7 8.7 4.3 8.7 13.0 4.3 48.0 - 2 5 9 2 1 6 5 31 1.5 2.9 7.4 13.2 2.9 2.9 2.9 1.5 8.8 7.4 46.0 MODERATE Risk Countries SDR . 50-74 : 68 13.0 13.0 3 4.3 3 65 3 62 4.3 54 4.3 4.3 13.0 4 4.3 50 13.013.0 5 . 8 35.0 35.0 35.0 | t t LOW Risk Countries SOR 50 49 HNIOW 49 1 42 4.3 4.3 4.3 HNIOW 46 38 1 - I - 31 J. 4.3 4 17.4 16 1 4.4 4.4 2.9 1.5 7 13.2 Syria Syria Turkey ~ - 112 1.5 112 2 1.5 2.9 2.9 1 SDRs from Doll K Worldwide distribution of gastrointestinal cancer Nat Can Inst Mono 25 1968 . + SDR = Standardized Death Rate per _ 100,000 JACKSON 025614 174 TABLE 33 AGE- AND SPECIFIC DEATH RATES PER 100,000 FROM STOMACH CANCER ANG ANG MALES AND RTSK CATEGORIZATIONS BASED ON THESE RATES Country Austria Czechoslovakia England France Germany Greece Hungary Italy Poland Rumania Sweden Scotland Scotland Yugoslavia Mortality per 100,000 65-74 75+ 251 263 154 117 204 100 274 217 274 221 121 149 165 439 432 247 242 429 160 464 - 379 342 261 286 227 215 CATEGORIZATION BASED ON RATES FOR YOUNGER MALES,, High Risk % Country Cases Hungary Poland Czechoslovakia Austria Rumania 10.5 10.5 5.3 15.8 - Totals 42 % Controls Moderate Risk % % Country Cases Controls 16.7 3.7 9.3 11.1 5.6 Italy Germany Yugoslavia Wales 15.8 5.3 5.3 15.8 _ 31.5 5.6 5.6 9.3 _ 46 42 52 Low Risk % % Country Cases Controls Scotland Sweden France 5.3 - 5.3 - 1.8 - Greece 5.3 I 16 N Hungary Austria Czechoslovakia Germany Italy Totals CATEGORIZATION BASED ON RATES FOR OLDER MALES 10.5 15.8 5.3 5.3 15.8 53 16.7 11.1 9.3 5.6 31.5 74 Poland Sweden Rumania Wales 10.5 - 15.8 __ 26 3.7 1.8 5.6 9.3 _. 20 France Scotland Yugoslavia Greece 5.3 5.3 5.3 5.3 _ 21 - 5.6 - 6 1 Mortality Rates from World Health Statistics Annual I Vital Statistics and Causes of Death World Health Organization Geneva 1976 JACKSON 025615 175 TABLE 34 BIRTHPLACE OF SUBJECTS ACCORDING TO RISK CATEGORIZATION AND CONFIRMATION STATUS High Risk Moderate Risk Low Risk Totals Number 9 6 2 17 Per Cent 53 35 12 100 Unconfirmed Cases N = 38 Number High Risk 2 Moderate Risk 2 Low Risk 2 Totals 6 Per Cent 33 33 33 100 Matched Controls 17 40 19 ' 42 45 15 100 Matched Controls 15 58 7 .27 4 15 26 100 JACKSON 025616 176 TABLE 35 BIRTHPLACE OF NATIVE AMERICAN SUBJECTS BY PATHOLOGY CATEGORIZATICATOEGORINZATION OF TATES,, 1 High Risk - Death Rate > 16 per 100,000 Conn Ill Maine Mass Mich Minn NH NJ NM NY ND Pean RI SD Wis 2 Moderate Risk ~ Death Rate 13 and 1.6 Cal Col Va Wy Idaho lowa La Md . Mont Neb New Ohio Or Utah Ver Wash 3. Low Risk - Death Rate < 13 All other states and D.C. except Alaska and Hawaii Confirmed Cases and Matched Controls High Risk States Moderate Risk Low Risk Totals Cases Frequency Per Cent 1.2 25 . 21 45 14 30 47 100 Controls Frequency Per Cent 36 102 69 - 207 17 49 33 100 Unconfirmed Cases and Matched Controls High Risk Moderate Risk Low Risk Totals Cases Frequency Per Cent 7 17 7 31 23 54 23 100 Controls Frequency Per Cent 23 ; 49 45 117 20 42 38 100 1 Based on white male death rates in Caucer by County 5 JACKSON 025617 177 TABLE 36 UNDERLYING CAUSES OF DEATH 1. ALL CASES UCD ICDA Code Malig Neo Esoph Malig Neo Stom Malig Neo Sm Int Malig Neo Skin Lymph & Hema Neo Isch Heart Disease Sympt Heart Dis Dis of Art Sen and Def Dis Unknown Totals 150 151 152 173 202 410-414 427 447 796 - Frequency 1 89 1 1 1 5 1 1 1 102 Per Cent 1.01 87.2 1.0 1.0 1.0 5.0 1.0 1.0 1.0 1.0 100.0 JACKSON 025618 TABLE 37 UNDERLYING CAUSES OF DEATH 2. ALL CONTROLS 178 UCD - Lymph & Hemat Neo Diabetes Mellitus Hypertensive Disease Ischemic Hrt Dis ICDA Codes 202 250 404 410-414 Sympt Hrt Dis def Hrt Dis Cerebrovasc Dis Dis of Art 427 429 430-438 440-448 Phleb and Thrombo 451 Pneumonia 480-486 Emphysema 492 Othr Dis of Resp Sys 510-519 Acute Append 540 Intest Obs 560 Othr Liver Dis 573 Chron Nephritis 582 Def Fract of Humerus Open Back Wound Total 796 812 876 ~ Frequency 1 3 1 56 1 1 11 5 1 N. 5 2 1 I I 1 1 1 1 96 Per Cent 1.0 3.1 1.0 : 58.3 1.0 1.0 11.5 5.2 1.0 2.1 5.2- 2.1 1.0 1.0 1.0 1.0 1.0 1.0 1.0 100.0 Alive - 290 These subjects were still alive on 01/01/76 the termination date of the study JACKSON 025619 179 TABLE 38 SUMMARY OF PATHOLOGY FINDINGS Method of Diagnosis Autopsy Histology Radiology Surgery Clinical Death Certificate Only Totals Number 6 50 4 4 2 38 102 Per Cent 6 49 4 4 . 2 37 100 Type of Cancer Adenocarcinoma NOS Un- or poorly differentiated Adenoca Well differentiated Adenoca Carcinoma or Cancer NOS Other Unknown Totals Site Within Stomach Upper Third cardia and fundus Mid Third body Lower Third pylorus . Other Combination Unknown Totals Number 40 10 5 9 5 33 102 Number 10 7 18 62 102 Per Cent 39 10 5 9 5 32 100 Per Cent 10 7 | 18 5 61 102 BATO JACKSON 025620 180 TABLE 39 FIRST CALENDAR YEAR IN THE INDUSTRY 1. A COMPANY CASES VS. CONTROLS Calendar Year 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 Totals Cases Frequency Per Cent 11 21 2 17 1 52 21.1 40.4 3.8 32.7 1.9 100.0 Controls ' Frequency Per Cent 23 78 22 70 4 197 11.7 39.6 11.1 35.5 2.0 100.0 JACKSON 025621 181 TABLE 40 FIRST CALENDAR YEAR IN THE INDUSTRY 2 COMPANY B CASES VS. CONTROLS Calendar Year 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 Totals Cases Frequency Per Cent . 21 16 1 11 1 50 42 32 2.0 22.0 2.0 100.0 Controls Frequency Per Cent 73 71 5 40 7 _ 196 37.3 36.2 2.6 20.4 3.6 100.0 JACKSON 025622 182 TABLE 41 FIRST CALENDAR YEAR IN THE INDUSTRY 3 ALL CASES VS. ALL CONTROLS Calendar Year 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 Totals Cases Frequency Per Cent 32 37 3 28 2 102 31.4 36.3 3.0 27.5 2.0 100.0 Controls Frequency Per Cent 96 149 27 110 11 121 24.4 37.9 6.9 28.0 2.8 - 100.0 JACKSON 025623 TABLE 42 PAIRED TESTS FOR YEAR OF HIKE OVERALL COMPARISON COMPARISON Mean Cases Controls S.D. Cases Controls Cases Controls Tailed Probability All Cases vs. All Controls 1928.0 1929.0 11.0 8.7 1.1 0.9 0.242 All Cases vs. All ARS Controls 1928.0 1928.8 11.0 9.5 1.1 0.9 0.44 All Cases vs. All ARSD Controls 1928.0 1929.2 11.0 10.3 1.1 1.0 0.07 Company A = All Cases vs. All Controls 1930.1 1931.6 10.3 7.2 1.4 1.0 - Cases vs. ARS Controls 1930.1 1931.3 10.3 8.5 1.4 1.2 ~ Cases vs. ARSD Controls 1930.1 1932.3 10.3 8.8 1.4 1.2 0.17 0.48 0.015 Company B All Cases vs. All Controls 1925.8 1926.1 11.3 9.3 1.6 1.3 0.78 - Cases vs. ARS Controls 1925.8 1926.3 11.3 9.8 1.6 1.4 0.72 - Cases vs. ARSD Controls 1925.8 1926.1 11.3 10.9 1.6 1.5 0.73 Confirmed Cases All Cases vs. All Controla 1928.4 1929.6 10,4 9.3 1.3 1.2 0.20 COMP A Cases vs. Controls 1930.4 1932.5 9.5 7.3 1.7 1.3 0.10 COMP B Cases vs. Controls 1926.3 1926.5 11.1 10.3 2.0 1.8 0.87 183 JACKSON 025624 Unconfirmed Cases All Cases vs. All Controls COMP A Cases vs. Controls COMB B Cases vs. Controls TABLE 42 cont'd PAIRED TESTS FOR YEAR OF HIRE - cont p 2 Mean Cases Controls Cases Controls Cases S.E. S.E. Controls Two Tailed Tailed Probability 1927.3 1927.8 12.0 7.7 2.0 1.2 0.72 1929.6 1930.3 12.0 7.0 2.7 1.6 0.78 1925.0 1925.4 12.0 7.8 2.8 1.8 0.82 184 JACKSON 025625 185 TABLE 43 FIRST CALENDAR YEAR IN THE INDUSTRY BY PATHOLOGY I. . CONFIRMED CASES VS. MATCHED CONTROL CONFIRMED CASES AND MATCHED CONTROLS Calendar Year Cases Frequency Per Cent Controls Frequency Per Cent 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 Totals 16 27 3 17 1 64 25 42 5 27 In 100 62 ; 86 20 72 _ 9 249 . 25 34 8 29 4 100 JACKSON 025626 186 TABLE 44 FIRST CALENDAR YEAR IN THE INDUSTRY BY PATHOLOGY II UNCONFIRMED CASES VS. MATCHED CONTROLS Calendar Year 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 Totals Cases Frequency Per Cent 16 10 - 42 26 1 11 1 38 30 3 100 Controls Frequency Per Cent 34 63 . 7 38 2 144 24 44 5 26 1 100 JACKSON 025627 187 TABLE 45 CHARACTERISTICS OF COMPANY A SUBJECTS WHO ENTERED THE THE INDUSTRY DURING AND AFTER THE WITH APPROPRIATE FIRST DECADE CONTROLS CONTROLS COMPARED COMPARED Decade 1910-1919 1. Cases Mean Median Total Duration 117 420 Age at Hire Age at Death 2. ARSD Controls 4.0 4.3 25.2 77.0 Total Duration 38 126 336 Age at Hire Decade Decade 1920 + 2.6 8.5 33 1 Cases Total Duration 325 120 332 Age at ire Age at Death 2 ARSD Controls 32.9 68.7 12.1 9.7 31 69.7 Total Duration 321 17 107 354 Age at Hire 33.5 1.6 10.5 31.1 Mode 167 22 73 88 24.5 318 19 68 150 22.0 JACKSON 025628 _ TABLE 46 TESTS FOR CROUP HIRED DURING FIRST DECADE BY PATHOLOGY ; COMPARISON Comp A Cases vs. ARSD Controls Mean Cases Controls S.D. Cases Controls S.E. Cases Controls Tailed Probability 1 Total Duration 338 320 121 110 17 15 2 Age at Hire 31.4 33.5 11.2 10.0 1.6 1.4 3. Year of Hire 1930.1 1932.3 10.3 8.8 1.4 1.2 First Decade Cases vs. ARSD Controls 0.09 0.02 0.015 1. Total Duration . 2 Age at Hire 385 317 117 25.5 34.4 4.0 3 Year of Hire 1917.1 1926.9 ~ 2.0 Confirmed 126 35 38 8.5 1.2 2.6 7.2 0.6 2.2 0.078 0.006 0.002 1 Total Duration 2. Age at Hire 3. Year of Hire 326 280 150 24.6 35.0 1.7 1916.8 1929.1 2.7 116 67 52 4.8 0.7 2.1 ' 3.4 1.2 1.5 0.396 0.014 0.007 . Total Duration 434 347 . 53 137 22 56 . 0.156 2 Age at Hire ; 26.3 33.9 5.4 11.1 2.2 4.6 0.145 188 3 Year of Hire 1917.3 1925.1 1.5 9.3 0.6 3.8 0.095 188 JACKSON 025629 TABLE 46 cont'd TESTS FOR GROUP HIRED DURING FIRST DECADE BY PATHOLOGY - cont p 2 Mean Mean Cases Controls S.D. Cases Controls S.E. Cases Controls Tailed Tailed Probability Later Decade Cases vs ARSD Controls 1 Total Duration 325 321 120 107 19 17 2 Age at Hire 32.9 33.5 12.1 10.6 1.9 1,7 : 3 Year of Hire 1933.6 1934.0 8.6 8.6 1.4 1.4 0.643 0.365 0.557 189 BATO JACKSON 025630 190 TABLE 47 TOTAL DURATION OF EMPLOYMENT IN MONTHS IN ACTIVE OTS 1 COMPANY A CASES VS. CONTROLS Total Duration 149 150-199 200-249 250-299 300-349 350-399 400-449 450-499 499 Totals Cases Frequency Per Cent 3 8 4 4 6 6 9 9 3 52 5.7 15.4 7.7 7.7 11.5 11.5 17.3 17.3 5.8 100.0 Controls Frequency Per Cent 16 14 17 31 32 35 23 20 9 197 8.0 7.1 8.6 15.7 16.2 17.8 11.7 10.1 4.6 ~ 100.0 Mean Median . S.D. S.E. 337.7 351.5 120.9 16.8 325.0 331.4 110.9 7.9 JACKSON 025631 191 TABLE 48 TOTAL DURATION OF EMPLOYMENT IN MONTHS IN ACTIVE OTS 2 COMPANY B CASES VS. CONTROLS Total Duration < 149 150-199 200-249 250-299 300-349 350-399 400-449 450-499 > 499 Totals Cases Frequency Per Cent | 1 2.0 3 6.0 3 6.0 8 16.0 3 6.0 5 10.0 12 24.0 7 14.0 001 50 16.0 100.0 Controls Frequency Per Cent * 2 1.0 13 6.7 14 | 23 7.2 11.7 14 7.2 36 18.4 28 14.3 | 39 27 196 19.9 13.8 100.0 Mean Median S.D. . S.E. 380.0 406.5 | 113.2 16.0 380.5 394.5 112.1 8.0 JACKSON 025632 192 TABLE 49 TOTAL DURATION OF EMPLOYMENT IN MONTHS IN ACTIVE OTS 3. ALL CASES VS. ALL CONTROLS Total Duration , < 149 150-199 200-249 250-299 300-349 350-399 400-449 450-499 > 499 Totals Mean Median S.D. | S.E. Cases Frequency Per Cent 4 11 7 12 9 11 21 16 11 102 4.0 10.8 6.9 11.7 8.8 10.8 20.6 15.7 10.8 100.0 358.5 385.5 118.6 11.7 Controls Frequency Per Cent 18 | 4.7 27 31 6.9 7.9 54 46 . 71 13.8 11.7 18.1 51 59 36 393 13.0 15.0 9.2 100.0 352.7 7 369.3 114.8 5.8 JACKSON 025633 193 TABLE 50 TOTAL DURATION OF EMPLOYMENT IN ACTIVE OTS BY PATHOLOGY I. CONFIRMED CASES VS. MATCHED CONTROLS Total Duration 149 150-199 200-249 250-299 300-349 350-399 400-449 450-499 500+ Totals Cases Frequency Per Cent 1 2 8 13 4 6 8 13 6 9 3 5 14 22 12 19 8 13 64 100 Controls Frequency Per Cent 10 13 . 22 34 28 45 34 34 29 249 4 5 9 14 11 18 14 14 12 100 Mean 368 360 JACKSON 025634 194 TABLE 51 TOTAL DURATION OF EMPLOYMENT IN ACTIVE OTS BY PATHOLOGY II UNCONFIRMED CASES VS. MATCHED CONTROLS | Total Duration Cases Frequency Per Cent 149 150-199 200-249 250-299 300-349 350-399 400-449 450-499 500+ Totals : oo 3 3 3 4 3 8 7 4 lw 38 8 00 00 10 8 21 18 10 8 100 Controls Frequency Per Cent 8 14 - 9 20 18 26 17 6 10 6 14 282 282 12 25 17 7 144 5 100 Mean 342 342 JACKSON 025635 TABLE 52 PAIRED TESTS FOR DURATION IN ACTIVE OTS OVERALL COMPARISON Mean Cases Controls S.D. Cases Controls S.E. Cases Controls Tailed Probability All Cases vs. All Controls 358 353 118 76 12 8 0.54 All Cases vs. All ARS Controls 358 357 118 88 12 9 0.94 All Cases vs. All ARSD Controls 358 350 118 109 12 11 0.24 Company A All Cases vs. All Controls - Cases vs. ARS Controls - Cases vs. ARSD Controls 338 326 121 75 17 10 338 335 121 87 16 12 338 320 121 110 17 15 0.38 0.90 0.09 Company B All Cases vs. All Controls - Cases vs. ARS Controls - Cases vs. ARSD Controls 380 381 380 380 380 382 113 67 113 83 113 100 16 10 16 12 16 , 14 0.96 0.98 0.88 Confirmed Cases All Cases vs. All Controls 368 360 119 77 15 10 0.39 COMP A Cases vs. ~ Controls 344 331 124 28 22 13 0.35 195 195 COMP B Cases vs. Controls 394 391 111 69 20 12 0.82 JACKSON 025636 TABLE 52 cont'd PAIRED TESTS FOR DURATION IN ACTIVE OIs - cont pg 2 Uncomfirmed Cases All Cases vs. All Controls COMP A Cases vs. Controls COMP B Cases vs. Controls Mean _ Cases Controls S.D. Cases Controls S.E. Cases Controls Tailed Probability 342 341 327 319 358 364 117 74 118 79 117 62 19 12 222 18 27 14 0.96 0.76 0.81 196 JACKSON 025637 197 TABLE 53 AGE AT HIRE 2 COMPANY B CASES VS. CONTROLS Age at Hire 15-19 20-24 . 25-29 30-34 35-39 40-44 45-49 50-54 54 Totals Mean Median S.D. S.E. Cases Frequency Per Cent , 7 14.0 15 12 7 3 1 2 30.0 24.0 14.0 6.0 2.0 4.0 "fi 1 50 4.0 2.0 100.0 28.1 25.5 9.6 1.4 Controls Frequency Per Cent 21 61- 38 21 28 14 8 4 196 10.7 31.1 19.4 - 10.7 ' 14.3 7.1 4.1 2.0 0.5 100.0 28.8 26.6 9.0 0.6 JACKSON 025638 198 TABLE 54 AGE AT HIRE 1 COMPANY A CASES VS. CONTROLS Age at Hire Cases Frequency Per Cent | 15-19 6 11.5 20-24 12 23.1 25-29 11 21.2 30-34 35-39 40-44 45-49 7 3 . 4 4 13.5 5.8 7.7 7.7 50-54 4 7.7 54 Totals 1 1.9 52 100.0 Controls Frequency Per Cent 8 n 39 4.1 19.8 34 17.3 36 18.3 24 12.2 22 - 11.2 14 17.1 14 6 __ 197 17.1 3.0 100.0 Mean Median S.D. S.E. 31.4 27.5 11.2 1.6 33.4 31.6 10.4 0.7 JACKSON 025639 199 TABLE 55 AGE AT HIRE 3. ALL CASES VS. ALL CONTROLS Age at Hire 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 54 Totals Cases Frequency Per Cent | 13 12.7 27 26.5 23 22.5 14 13.7 6 5.9 5 4.9 .6 5.9 6 5.9 2 102 2.0 100.0 Controls Frequency Per Cent 29 100 - 72 57 52 36 22 18 7 393 7.4 25.4 18.3 14.5 13.2 9.2 5.6 4.6 1.8 100.0 Mean Median S.D. S.E. | 29.8 26.8 10.5 1.04 31.1 29.1 10.0 0.5 JACKSON 025640 ~ 200 TABLE 56 AGE AT HIRE BY PATHOLOGY CONFIRMED CASES VS. MATCHED CONTROLS . Age at Hire 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 54 Totals Cases Frequency Per Cent Controls Frequency Per Cent 8 15 19 6 4 4 4 3 1 64 12 23 30 9 6 6 6 5 100 20 61 - 43 34 35 30 13 11 2 249 8 24 17 14 14 12 5 4 | 100 JACKSON 025641 201 TABLE 57 II AGE AT HIRE BY PATHOLOGY UNCONFIRMED CASES VS. MATCHED CONTROLS Age at Hire 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55 Totals Cases Frequency Per Cent , . 5 13 12 32 4 11 8 21 2 5 1 3 2 5 3 8 1 3 38 Controls Frequency Per Cent 9 39 - 29 23 17 6 9 7 5 144 6 27 20 '16 12 4 6 S 3 Mean 30.2 31.1 JACKSON 025642 TABLE 58 PAIRED TESTS FOR AGE AT HIRE OVERALL COMPARISON Mean Cases Controls S.D Cases Controls S.E. Cases Controls Tailed Probability All Cases vs. All Controls 29.8 30.1 10.5 7.1 1.0 0.7 0.26 All Cases vs. All ARS Controls 29.8 30.5 10.5 7.8 1.0 0.8 0.49 All Cases vs. All ARSD Controls 29.8 31.0 10.5 9.2 1.0 0.9 0.07 Company A All Cases vs. All Controls 31.4 32.9 11.2 7.6 1.6 1.1 0.18 - Cases vs. ARS Controls 31.4 32.5 11.2 8.1 1.6 1.1 0.49 - Cases vs. ARSD Controls 31.4 33.5 11.2 10.0 1.6 1.4 0.02 Company B All Cases vs. All Controls 28.1 28.4 9.6 5.7 1.4 0.8 0.84 - Cases vs. ARS Controls 28.1 28.5 9.6 6.9 1.4 1.0 0.81 - Cases vs. ARSD Controls 28.1 28.5 9.6 7.6 1.4 1.0 0.73 Confirmed Cases , . All Cases vs. All Controls 29.6 30.6 10.2 7.3 1.3 0.9 0.24 A CasCeass es vs. Controls 31.1 33.0 11.4 7.4 2.0 1.3 0.14 B COMP Cases vs. Controls 28.0 28.1 8.8 6.4 1.6 1.1 0.88 202 JACKSON 025643 Unconfirmed Cases All Cases vs. All Controls COMP A Cases vs. Controls COMP B Cases vs. Controls TABLE 58 cont'd PAIRED TESTS FOR AGE AT HIRE - cont pg 2 Mean a Cases Controls S.D. Cases Controls S.E. Cases Controls TTaaiile lded Probability 30.2 30.8 11.2 6.8 1.8 1.1 0.69 31.9 32.8 11.3 8.1 2.6 1.9 0.68 28.4 28.7 11.1 4.7 2.5 0.89 203 . JACKSON 025644 ANALYSIS II OCCUPATIONAL EXPOSURE This second of two analysis chapters will present the findings relating specifically to occupational exposures and will utilize the OT and OTG System used previously in other OHSG studies Because the differences between cases and control in total duration of employment employment within active OTs were not statistically significant the analysis of control work histories will not be terminated at the point where the control attained the duration of the case as was stipulated during the earlier description of the design for this study Also because there was no statistically significant difference in duration within active OTS between ARS- and matched controls this dichotomy will be essentially ignored during the following discussion Most Representative OTGS Before the calculation of odds ratios for specific OTGs it would be useful to view the work history data in a general manner for which purpose Tables and 62 present the first and second most representative OTGs defineads those OTGs in which subjects spent the longest and second largest periods of time for both cases and controls in each company separately and combined Among Company A cases the five most representative OTGS are general services fabrication of tires and beads batch preparation curing and metal handling while controls had general services chemicals fabrication maintenance and three OTGs tied for fifth as their most representative OTGs Among Company B subjects the most representative OTGs for cases were Maintenance fabrication JACKSON 025645 205 general services and batch preparation while among controls the MROTGs were maintenance fabrication general services batch preparation and calendering Among Company A subjects therefore two of the five MROTGS were shared in common by cases and controls while among Company B subjects four were shared in common We then calculated odds ratios and chi squares for each of these OTGs and did not find any significant elevations for MROTGS and observed only two second MROTGS Batch Prep Inspection Prep Other in Company B with significantly elevated ORs Standard OTG Analyses The data were first examined according to the standard OTG classification used in numerous other OHSG reports a classification which includes twenty OTGS See Appendix I a for description of the industrial processes included in the OTGs which comprise them based on similar production processes and materials but not necessarily based on similar exposures For this reason analyses were also performed on specific OTGs created especially for the suspect | gastric carcinogens in this investigation The Odds Ratios for the standard OTGs were examined in numerous ways each addressing a different question with each examination being equivalent to stratification on the factor under examination a procedure which controls for confounding In all we examined the odds ratios according to control group ARS and ARSD pathology confirmation status year of hire birthplace age at hire and age at death in addition to the chemical specific analyses with many of these stratifications including four different exposure times JACKSON 025646 206 ever two years five years and ten years - each of which were compared with no exposure The results of these analyses are presented in Table6s3 to 95. It should be noted before examining these tables that one or two asterisks indicate the value tailed for a test of significance of the odds ratios being different from unity The statistic CHI is an approximate standard normal variate In addition the ninety percent confidence interval is presented along with the numbers of exposed and unexposed cases and controls Examination of the first seven tables which present the overall and company and control group specific odds ratios on an never exposure basis reveals not a single significantly elevated odds ratio although the ORs for calendering tuber and cure preparation were somewhat elevated for comparisons with one or the other control group within Company A. In Company B on the other hand the OR for calendering was twice statistically significantly lower than unity This first set of analyses of the data examining exposure on an never basis was followed by further examination on the basis of two five and ten years of exposure within each of these OTGs all compared with no exposure at all Each of these analyses was performed separately for each company Tables 70 to 75 and only one OR - Calendering among Company B subjects in the comparison of two years of exposure versus none - was statistically significantly elevated while Fabrication appeared elevated twice in this sequence of analyses When these analyses were further stratified according to pathology confirmation status the following significantly elevated ratios with a considerable number of exposed cases were observed among confirmed cases - metals for two or more years in Company A OR = 2.30 with 14 JACKSON 025647 207 exposed cases Fabrication for both two OR = 3.1 with 8 exposed cases and five or more years OR = 3.5 with 7 or more cases in Company A and among unconfirmed cases - none We then stratified subjects according to their pathology confirma- tion status Tables 76 and 77 analyzing separately confirmed and unconfirmed cases and their matched controls on the assumption that there may be contamination by stomach cancer cases among the unconfirmed group This analysis indicated two elevated ORs - for curing and Prep -Other both of which occurred among the unconfirmed cases The next analysis was based on the age at hire of the case which was dichotomized according to whether the case began work in the industry here was either before or after the to separate those for whom age the of twenty The attempt | exposure began at a young age compared with those whose exposure began at an older age and the results of these analyses are presented in Tables 78 to 83. Among the younger group overall we observed elevated odds ratios for Fabrication and Cure Preparation with cases in Company B statistically associated for the overall elevation Within this company the odds ratios for Curing and Prep - Other were slightly although not statistically significantly elevated as well Among Company A subjects exposed at younger ages on the other hand there was a significant elevation for Calendering and significantly elevated ORs for Tuber Cureprep Curing and Chemicals although the number of exposed cases in all of these instances was small Similar analyses for individuals whose first exposure occurred at an older age did not reveal any elevated ratios at all Since the groupings for age at hire JACKSON 025648 208 are based on the values for the case when the matched analyses are conducted the age at hire for the controls can take on any value Accordingly we dissolved the matching and calculated unmatched odds ratios but did not observe any significant elevations We next looked separately at cases and their matched controls who died before and after the age of sixty Tables 84-87 since it has been postulated that these may be two distinct entities with different causal mechanisms one among young cases and the other among the older ones There were a few elevated ORs among those who died at younger ages - Curing General Service Final Inspection Calendering and Chemicals - in Company A and none among this same group in Company B. Among those who died at older ages in Company A on the other hand there were no significant elevations although Tubers was somewhat high Among subjects in Company B once again we found nothing of interest For the analysis based on birthplace and decade of hire we dissolved the matching for reasons stated earlier For the former the only significant elevated odds ratios were found for the Calendering and Tuber OTGs among native born employees at Company A where the ratios were each 2.6 with 9 and 11 exposed cases respectively For the decade of hire analysis separate odds ratios and CHIS were calculated for each of the OTGs for five decades 1910-1959 in each Company separately The only significant finding was that for both Company A and B the odds ratios were high 5.6 and 4.0 for cases in Fabrication who entered the industry during the first decade For the 11 Company A cases in this group5 were exposed while 15 of 21 cases in Company B had been employed in the Fabrication OTG BATO JACKSON 025649 209 The next series of analyses examined the odds ratios for those OTs judged to have potential for exposures to four specific chemicals or classes - polycyclic hydrocarbons nitrosamines carbon black and asbestos or asbestos containing talc Three environmental scientists at OHSG whose responsibilities include detailed studies of the occupational environment of workers at the two companies under study were asked to independently rate each of the OTs accordinthge level exposure moderate low or none - for each of the four chemicals under study Since this was an ordinal assessment high moderate low or none the numerical values of three two one and zero were assigned to these levels and an average of all three judges was then calculated for each OT see Appendix III for inter- observer reliability After these means had been calculated off points for each type of exposure were established after which we had groups of OTs considered to have high moderate low or no worker potential for exposure to the four chemicals under investigation and a listing of these groups is presented in Table 98. Because those judgements were subjective we first analyzed the high groups compared with all others and we then combined the high and moderate group and compared those against the remainder the former always being considered exposed while the latter was considered unexposed Each of those analyses was performed in addition for the four exposure groups of never and two five and ten years versus none and the results are presented in Tables 88 to 95. Examination of these results _ indicates the following no elevated odds ratios in either company for the aromatic hydrocarbons a slightly but significantly elevated OR for Company B subjects exposed for five or more years to high levels of JACKSON 025650 210 nitrosamines OR = 1.7 with significant elevations for Company B subjects as well slightly elevated ratios for exposure to carbon black among Company A and B subjects although none were statistically significant and statistically significantly elevated odds ratios for Company A subjects exposed to high or moderate levels of asbestos while Company B subjects had significantly low ratios for exposure to this same chemical These low ratios among Company B subjects however were found only for the combined high or moderate exposures When moderate exposure was compared against all others including high we observed slightly elevated ORs even in Company B. When these chemical specific analyses were further stratified according to pathology there were no significantly high ratios among confirmed cases in either company for any of the chemicals Among unconfirmed cases on the other hand we found high odds ratios for numerous combinations of asbestos exposure as well as for numerous time and exposure categories of polycyclic hydrocarbons nitrosamines and carbon black with all significant elevations occurring within Company A. It should be kept in mind when considering these findings however that there were only sixteen 31 percent unconfirmed cases among the fifty total cases in Company A. Because of the finding of elevated odds ratios for asbestos exposure among Company A subjects and because of the large number of potentially exposed subjects 27 additional attention was directed towards this category We next performed a paired Test and found that Company A cases averaged approximately four years 49 months in OTs judged to have potentially high asbestos exposure while their matched controls averaged half that exposure time 25 months Using JACKSON 025651 271 a tailed probability test of significance this difference was statistically significant at the five percent level A similar analysis among Company B subjects revealed no significant differences between cases and comparands as we expected from the absence of significant odds ratios for this chemical although subjects in this company spent slightly more time in high asbestos exposure OTs eighty- four months for cases and eighty for controls than did the subjects ; in Company A. After this finding two groups of Company A cases were further - examined - those who entered before and after 1920 since we had earlier found a disproportionate number of cases among the former Paired Tests revealed that the eleven early entry cases averaged 82 months in areas with potentially high asbestos exposure with their matched controls averaging 24.5 months while the respective findings for the 41 later entry cases and their matched controls were 40 and 26 months with neither of these differences achieving statistical significance When exposure differences among the early entry group was further analyzed separately for ARS- and matched controls the cases again averaged 82 months of course while controls averaged 35 and 12 months respectively The overall significant difference in potential asbestos exposure therefore appears to be statistically accounted for by the differences between early entry cases and their matched controls These analyses were further pursued by the calculation of unmatched odds ratios for potential asbestos exposure for each decade of entry and the results of these calculations are presented in Table 96 , from which we see that only for the second decade was the odds JACKSON 025652 212 ratio significantly elevated although there was a significant elevation for the first decade as well Of the 32 cases who entered during the first decade fully half had been exposed with 30 of the 73 controls having similar experiences while 7 of the twenty cases and 21 of the 74 controls who entered in 1930 or later had worked in areas with potentially high asbestos exposures It is interesting to note here that the similar matched analysis described elsewhere in this chapter which utilized standard OTGS did not indicate any consistent excesses Because these earlier entry odds ratios were found to be elevated we next examined the distributions of total time spent in potentially high asbestos OTs only for those individuals who had ever worked in these areas and who entered the industry before 1930. The cases in this group averaged 81.3 months in these OTs while controls had a mean of 66.8 For subjects entering in 1930 or later the mean duration in potentially high asbestos exposure areas was 48.9 months while controls averaged 70.8 The final analysis in this series examined the odds ratios and related factors for different latency periods beginning with 40 years These latency periods are calculated by subtracting the year of hire from the year of death although the reader should recall that most controls were still alive on 1/1/76 although this date was used as the year of death in these calculations Based on this determination of latency period we observe a statistically significantly elevated odds ratio of 2.24 for potential asbestos exposure among cases compared with controls who had the same potential latency period The results of this analysis are presented in Table 97. The results of both the potential latency and calendar year of entry analyses are summarized in JACKSON 025653 a 213 Table 99 This analysis was followed by one which examined the calendar periods during which the two groups in Company A actually worked in these high risk OTs and the frequency distributions revealed that approximately fifty percent of the case instances of employment within these OTs had occurred by the year 1940 while only thirty percent of the control instances had occurred by this date Although this finding indicates that the exposures occurred well before the disease we further examined the Test for cases who died before and after the age of seventy Since retirement is mandatory by the age of sixty- five any exposures among those who died after the age of seventy must have occurred at least five years prior to death a time frame which can be considered well within the latency period Among Company A | subjects the results indicate that the difference in asbestos exposure between cases and controls was even greater among those dying at older ages fifty versus twenty months a finding which strengthens ourconfidence that exposures may have occurred enough earlier than the disease to have had an impact with a reasonably long latency period JACKSON 025654 214 TABLE 59 MOST REPRESENTATIVE OTGs MROTGs FOR COMPANY A OTG CONTROLS CUM % % N x x CASES 7 CUM OR % % N GENERAL SERVICE | 15 15 0.01 1.0 15 8 CHEMICALS 13 5.10 17 FABRICATION 10 38 19 || 0.64 1.5 13 30 7 MAINTENANCE CALENDER 10 48 0.77 0.6 | 6 36 3 7 14 0.02 8 44 CURING MILLING METAL HANDLING BATCH PREP 6 7 | 6 S INSPECTION 5 RECLAIM 4 SHIP & REC 3 TUBER 3 FIN INSPECT 1 TUBES FABRIC 1 UNKNOWN 1 61 13 | 0.56 68 14 0.73 74 1.11 79 10 1.50 84 10 |] 0.04 88 9 | 0.05 222 6 | 0.08 222 6 || 0.88 222 3 1 96 3 1.13 97 3 I 1.5 | 9 0.5 | 4 1.8 |] 9 2.0 || 9 1.1 6 0.8 4 1.3 4 2.0 | en - - 2.6 || 4 1 I 53 57 66 5 75 5 81 3 85 89 95 3 95 - 99 2 99 - CURE PREP 1 98 2 - - - 99 = OTHER RUBBER - 98 - - - - 99 = PLASTICS PREP OTHER - 98 1| - - 98 - - -| - -- 22 = 99 - SALARY - 1 E 98 - ~ 99 = TOTALS 98 197 | 99 52 JACKSON 025655 215 TABLE 60 SECOND MOST REPRESENTATIVE OTGS FOR COMPANY A OTG CONTROLS CUM % % N x x GENERAL SERVICE CHEMICALS | 9 00 MAINTENANCE 00 UNKNOWN 00 BATCH PREP 6 MILLING 6 CALENDER 4 CURING 4 RECLAIM 4 TUBES 4 CURE PREP 4 FABRICATION 4 INSPECTION 4 SHIP & REC 4 METAL HANDLING 2 TUBER 2 OTHER RUBBER 1 PREP OTHER I FINAL INSPECT - PLASTICS - SALARY - NO 2nd MROT 15 9 17 25 33 39 45 49 53 57 61 65 69 73 77 79 81 82 83 = - 1 98 17 || 0.46 16 2.5 16 | 0.01 16 || 0.01 12 || 0.39 12 || 0.39 | 9 0.81 9 || 0.13 9 I 9 f} 0.13 8 || 0.28 8 || 4.34 8 0.54 I} 8 0.54 5 ll 0.26 4 1 0.58 1 1 1 ~ -i - -~f - ~ |} 29 | J TOTALS 98 197 CASES CUM OR % % N 0.6 6 2.1 15 0.9 7 0.9 7 0.6 4 0.6 4 1.7 7 1.3 6 i - 1.3 6 1.4 3.1 0.5 0.5 6 f 11 2 2 1.5 4 1.9 4 1 - I - - - 1 - 1 1 - 7 6 21 28 35 39 43 50 56 56 62 68 = 79 81 83 87 91 91 91 91 91 91 98 3 8 4 4 2 2 4 3 - 3 3 6 .1 1 2 2 - ~ - ~. 4 98 52 JACKSON 025656 216 TABLE 61 MOST REPRESENTATIVE OTGS FOR COMPANY B OTG CONTROLS CUM % N x x MAINTENANCE 19 FABRICATION 11 GENERAL SERVICE BATCH PREP 9 CALENDER 9 19 38 21 41 21 50 18 || 0.05 sf 0.43 | 0.07 | 0.03 59 18 0.52 CURING TUBES CURE PREP INSPECTION 6 65 6 71 71 5 76 12 | 1.35 12 || 0.33 1 CO 9 0.03 PREP OTHER 5 81 10 | 0.06 TUBER 1 81 1 - PLASTICS SHIP & REC 4 85 4 89 8 | 0.01 7 0.60 UNKNOWN 4 93 7 0.02 FINAL INSPECT 1 94 METAL HANDLING | 1 95 MILLING 1 96 3 fl 1.22 3 0.05 2 | 0.3 OTHER RUBBER 1 97 3 - CHEMICALS RECLAIM SALARY 1 97 97 - 97 1 1 | 1 1 | - 1 TOTALS | 196 7 OR | % 0.9 | 18 14 CASES CUM % N 18 9 32 1.1 1.1 0.6 ~ 0.3 | 12 } 10 | 6 2 44 6 54 ~~ 5 60 3 228 1 0.6 4.0 4 228 2 2 228 1 0.9 4 72 2 1.2 6 78 3 - ~ 78 - 1.0 1.7 1.1 2.7 1.3 2.0 4 6 4 4 | 2 2 82 88 3 92 2 96 2 98 1 100 1 1 - 100 - 1 - 100 1 1 -~ 100 I I - 100 = | 100 50 JACKSON 025657 217 TABLE 62 SECOND MOST REPRESENTATIVE OTGS FOR COMPANY B OTG CONTROLS CUM % % x N CASES CUM OR | % % N GENERAL SERVICE 15 TUBES 10 CALENDER 8 MAINTENANCE 7 BATCH PREP 6 FINAL INSPECT 5 CURE PREP 4 FABRICATION 4 INSPECTION 4 UNKNOWN 4 CURING 3 PLASTICS 3 PREP OTHER 2 RECLAIM 2 SHIP & REC 3 CHEMICALS 1 MILLING 2 OTHER RUBBER 2 TUBER 1 METAL HANDLING | 1 SALARY - NO 2nd MROT 13 15 25 33 40 46 51 55 59 63 67 70 73 75 77 80 81 83 85 86 87 87 100 29 ff 2.7 20 3.4 0.4 7 0.2 f 2 16 - - I - 13 | 0.03 0.9 14 3 11 4.13 2.7 14 28 7 9 0.17 1.3 34 3 7 - - - 34 - 8 1.32 2.1 7 00 42 4 7 5.65 3.7 12 54 6 7 | 0.60 1.7 6 60 3 6 - - - 60 1 6| - I 88 - 5 8.33 5.2 12 72 6 5 | 0.05 6 - 0.8 - 2 74 1 ~ 74 - 3 - 4 - I - 74 - I - 74 - 4 2.26 3.0 | 6 80 3 3 | 1.22 2.7 4 84 2 2 - - - 84 - -| - 25 1 - - 84 1 1 16 100 8 TOTALS 100 196 100 50 JACKSON 025658 TABLE 63 ODDS RATIOS FOR ALL CASES COMPARED WITH ALL THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1. BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION . FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16. PREP - OTHER 17 FINAL INSPECTION 18. METAL HANDLING 19 OTHER RUBBER 20. SALARY 21 UNKNOWN CASES EXP UNEXP 27 75 8 94 25 77 17 85 36 66 25 77 17 85 30 72 22 80 28 74 42 60 16 86 10 2 4 2 5 5 24 25 15 87 28 74 8 94 4 98 15 87 = CONTROLS EXP UNEXP ; CHI VALUE OR 81 312 1.42 1.48 CONFIDENCE INTERVAL LOW HIGH 0.94 2.33 49 344 1.48 0.52 0.25 1.08 114 279 0.99 0.78 0.52 1.15 59 334 0.38 1.12 0.70 1.79 119 274 0.99 1.29 0.84 1.97 103 290 0.33 0.92 0.60 1.40 49 344 - 1.22 99 294 0.75 90 303 0.25 90 303 0.90 1.43 0.88 2.32 1.20 0.81 1.77 0.92 0.56 1.53 1.27 0.82 1.97 163 225 0.18 58 235 0.27 48 345 0.68 0.96 0.66 1.40 1.09 0.64 1.81 0.79 0.44 1,41 20 373 0.48 0.76 0.30 1.93 36 357 1.47 0.44 0.18 1.10 67 326 1.42 1.46 0.94 2.28 56 337 0.11 1.04 0.60 1.79 99 294 0.49 47 346 0.18 1.16 0.71 1.87 0.60 0.29 1.22 18 375 0.22 0.88 0.35 2.25 65 328 0.14 0.95 0.55 | 1.66 *** *** probability . ** /0.05 probability /0.01 218 JACKSON 025659 TABLE 64 ODUS RATIOS FOR COMPANY A CASES COMPARED WITH ALL THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG BATCH PREPARATION MILLING CALENDER TUBER FABRICATION TUBES -FABRICATION CURE PREPARATION CURINC INSPECTION MAINTENANCE GENERAL SERVICE SHIP & REC RECLAIM CHEMICALS PLASTICS 16 PREP ~ OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20 SALARY 21 UNKNOWN CASES EXP 11 41 7 45 13 39 11 41 15 37 12 40 8 44 14 38 922 922 25 27 46 FW 48 48 S 52 47 47 27 25 1 51 50 18 42 CONTROLS EXP UNEXP 28 169 28 169 37 160 24 173 42 155 36 151 18 179 44 153 33 264 43 154 76 121 24 173 29 168 7 190 0 197 10 187 12 185 85 112 3 194 16 181 33 164 CHI VALUE 1.461.46 0.96 0.27 1.58 1.58 1.58 1.12 0.81 0.81 1.481.48 1.19 0.07 1.33 1.37 1.12 0.80 1.23 0.18 0.98 0.55 CONFIDENCE INTERVAL LOW HIGH HIGH 0.92 3.96 3.96 0.35 0.78 2.53 2.53 0.98 0.82 0.73 0.93 0.74 0.37 3.36 3.36 2.81 2.81 2.56 2.56 4.25 2.40 1.71 0.68 0.86 10.42 0.21 0.84 0.77 0.61 0.87 0.17 0.15 0.62 H = probability /0.05 ** * probability 0.01 . 219 JACKSON 025660 TABLE 65 ODDS RATIOS FOR COMPANY A CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1 BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION 6 TUBES -FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13. RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP D OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19. OTHER RUBBER 7 20 SALARY 21 UNKNOWN CASES EXP UNEXP 11 41 CONTROLS EXP 13 ? 45 11 90 13 39 20 81 11 41 14 87 15 37 24 77 12 40 21 80 8 44 10 91 14 38 25 76 7 45 19 82 13 w 26 75 25 N 38 63 6 2 14 87 4 2 18 83 4 2 0 52 3 98 0 101 5 47 S. 96 5 47 6 95 27 25 37 64 1 51 2 99 .2 50 > 92 3 43 15 86 CHI VALUE OR 1.64 2.50 CONFIDENCE INTERVAL LOW 1.00 0.49 1.31 0.53 0.65 1.31 0.66 2.60 1.00 1.48 0.78 2.83 0.59 1.27 0.65 2.50 0.35 1.16 0.58 2.30 1.06 0.39 0.75 1.75 0.73 4.17 1.17 0.61 2.22 0.69 0.30 1.56 0.00 1.18 0.33 1.66 1.44 1.00 1.54 0.85 0.42 3.50 += 0.85 0.37 0.17 0.84 --- 2.80 1.93 0.99 14.60 --- 0.88 0.63 --- 1.67 1.50 e-- 0.64 0.52 --- 4.32 4.31 1.82 1.86 1.06 3.26 0.20 0.75 0.07 7.99 1.12 0.40 0.38 0.09 1.59 1.30 0.54 3.15 * ' probability /0.05 *** *** probability 0.01 . 220 JACKSON 025661 TABLE 66 ODDS RATIOS FOR COMPANY A CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1 BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION 6 FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP - OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20 SALARY 211 UNKNOWN CASES EXP UNEXP 11 41 7 45 13 39 11 41 15 37 12 40 8 44 14 38 7 45 13 39 25 26 6 26 4 48 + 48 0 52 5 47 5 47 27 25 1 51 2 50 9 43 CONTROLS EXP UNEXP 15 81 17 79 17 79 10 86 18 73 15 81 & 88 19 77 14 82 17 79 38 58 10 86 11 85 4 92 96 91 90 48 48 L 95 7 89 18 78 CHI VALUE 1.02 . 0.94 1.05 1.82 1.41 1.26 1.46 1.20 . 0.08 1.18 1.02 0.20 0.68 0.93 --- OR 1.59 0.56 1.50 2.38 1.83 1.80 2.25 1.65 0.96 1.67 1.44 1.13 0.65 1.88 --- CONFIDENCE INTERVAL LOW HIGH 0.76 3.35 0.20 1.54 0.80 2.S3 1.09 5.19 0.90 3.72 0.53 3.88 0.90 5.60 0.83 3.30 0.42 2.18 0.82 3.40 0.80 2.57 0.41 3.13 0.23 1.83 0.61 5.72 = --~ 1.34 2.25 0.83 6.11 0.78 1.70 0.55 5.22 0.48 1.19 0.65 2.17 0.50 2.00 0.20 19.56 0.71 0.57 0.16 2.10 0.17 0.92 0.43 2.00 ** ** probability /0.05 ** = probability 50.01 . 221 JACKSON 025662 TABL6E7 ODDS RATIOS FOR COMPANY B CASES COMPARED WITH ALL THEIR MATCHED CONTROLS EXPOSURE = NEVER : CASES CONTROLS CONFIDENCE INTERVAL 1 BATCH PREPARATION 16 2 MILLING 1 3 CALENDER 12 4 TUBER 6 5 FABRICATION 21 6 TUBES FABRICATION 13 7 CURE PREPARATION 9 8 CURING 16 9 INSPECTION 15 10 MAINTENANCE 15 11 GENERAL SERVICE 18 12 SHIP & REC 10 _ 13 RECLAIM 14. CHEMICALS 15 PLASTICS 16 PREP - OTHER 19 10 17 FINAL INSPECTION 1 18 METAL HANDLING 7 19 OTHER RUBBER 20 SALARY 2 5 21. . UNKNOWN 34 53 49 22326786 38 22326786 44 22326786 29 22326786 37 22326786 41 22326786 34 22326786 35 22326786 35 50 32 90 40 34 44 19 50 25 45 2232 31 2232 40 2232 49 2232 43 44 48 2 45 32 143 0.68 175 1.94 119 2.21 161 0.98 119 0.34 129 1.14 165 0.38 137 0.26 139 0.09 146 0.66 106 1.47 162 0.41 177 0.46 171 1.84 160 1.47 172 1.03 152 0.40 176 1.51 1.31 194 1.50 164 0.44 1.27 0.71 2.27 0.19 0.04 0.77 * 0.44 0.24 0.81 0.64 0.30 1.36 1.13 0.63 2.02 0.67 0.37 1.19 1.16 0.62 2.18 1.09 0.64 1.84 1.04 0.54 2.01 1.30 0.68 2.49 0.61 0.35 1.06 1.17 0.62 2.23 1.26 0.56 2.84 0.00 -- -- 0.44 0.18 1.10 1.39 0.82 2.34 0.85 0.43 1.65 0.27 ' 0.07 1.12 0.54 0.25 1.17 4.00 0.79 0.87 18.29 0.27 1.59 * s probability /0.05 ** = probability 0.01 ; 222 JACKSON 025663 TABLE 68 ODDS RATIOS FOR COMPANY B CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE NEVER CONFIDENCE INTERVAL INTERVAL CASES HIGH OTG EXP CHI VALUE LOW BATCH PREPARATION 16 34 23 MILLING . 49 6 CALENDER 12 38 39 TUBER 6 44 19 FABRICATION 21 29 44 FABRICATION 13 CURE PREPARATION 9 37 3222222 41 3222222 CURING 16 34 3222222 INSPECTION 15 35 3222222 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 15 35 3222222 18 32 3222222 10 40 3222222 0 4 ^ 14 CHEMICALS 0 50 7 75 1.25 92 1.07 59 2.00 79 1.08 54 0.31 59 1.71 81 0.07 69 0.20 68 0.14- 76 0.94 48 1.71 78 0.07 " 90 0.78 91 1.87 0.85 3.38 3.38 3.38 0.06 1.31 1.31 0.20 0.80 1.3 0.80 0.26 1.32 1.32 0.50 1.61 1.6 1.61 0.28 0.98 0.98 2.13 0.50 2.13 2.13 0.57 0.45 0.75 0.32 0.50 0.61 15 PLASTICS 5 257 19 79 1.46 16 PREP - OTHER 19 31 30 68 0.80 17 FINAL INSPECTION 10 40 25 73 0.82 18 METAL HANDLING 1 49 89 1.65 19 OTHER RUBBER 7 43 71 1.77 20 SALARY 2 48 99 2.00 21 UNKNOWN 5 45 13 85 0.00 223 fi= probability 0.05 ** = probability /0.01 JACKSON 025664 TABLE 69 ODDS RATIOS FOR COMPANY B CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE NEVER CASES CONTROLS CONFIDENCE INTERVAL 1 BATCH PREPARATION 2 MILLING 3 CALENDER 42 TUBER 5 FABRICATION 6 TUBES -FABRICATION 7 CURE PREPARATION 9 CURING 9 INSPECTION 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP - OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20. SALARY 21 UNKNOWN 16 34 30 4 49 15 12 38 38 6 44 16 21 29 33 13 37 28 9 41 16 16 34 30 15 35 27 15 35 28 18 32 42 10 40 14 44 11 0 50 8 5 45 17 19 31 29 10 40 19 49 S 7 43 2 48 2 5 45 14 .68 0.125 83 2.38 60 1.96 82 0.64 65 0.98 70 0.35 82 0.62 68 0.18 71 0.34 70 0.21 56 0.89 84 0.93 87 0.09 90 1.73 81 1.41 69 1.06 79 0.17 93 1.00 81- 0.60 96 0.71 84 0.77 1.05 0.13 0.46 0.71 1.46 0.87 1.31 1.06 1.16 1.10 0.71 1.54 1.05 A 0.38 1.44 1.08 0.36 0.72 2.00 0.64 0.57 0.03 0.24 0.30 0.78 0.45 0.64 0.61 0.57 0.52 0.37 0.71 0.44 --- 0.12 0.82 0.51 0.07 0.29 0.40 0.25 1.94 0.53 0.88 1.68 2.74 1.67 2.68 1.85 2.34 2.36 1.34 3.35 2.52 1.18 2.52 2.32 1.92 1.78 10.03 1.66 * " probability /0.05 ** = probability /0.01 224 JACKSON 025665 TABLE 70 ODDS RATIOS FOR COMPANY A SUBJECTS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE 2 YEARS VS. NONE OTG 1 BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION 6. TUBES -FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION . 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP - OTHER 17 FINAL INSPECTION 18 METAL HANDLING ~~ 19 OTHER RUBBER - 20 SALARY 21 UNKNOWN CASES EXP UNEXP 7 41 4 45 8 39 8 41 11 37 8 40 6 44 8 38 S 45 7 39 12 27 4 46 3 48 2 48 0 523 2 523 0 523 16 25 0 51 50 4 43 CONTROLS EXP UNEXP 2 162 21 160 23 143 17 164 21 145 17 150 12 167 18 135 17 152 26 138 45 86 13 167 16 164 6 182 CHI VALUE OR 1.15 2.0 0.92 0.53 0.53 1.26 1.31 1.78 1.74 2.08 1.07 1.67 0.97 1.78 0.88 1.46 0.79 0.61 0.01 1.04 0.52 0.79 0.19 1.12 0.73 0.62 0.29 1.30 CONFIDENCE INTERVAL LOW HICH 0.74 5.40 0.17 1.63 0.61 2.62 0.86 3.65 1.04 4.15 0.76 3.65 0.67 4.77 0,72 2.95 0.22 1.70 0.51 2.12 * 0.38 1.60 0.43 2.92 0.21 1.85 0.30 5.67 0 197 ~- 2 175 1.24 5 169 1.15 43 92 1.28 2 191 0.71 10 181 0.33 17 152 0.33 --- 3.1 0.00 1.59 0.00 0.76 0.82 --- 0.69 - - 0.87 oo 0.19 0.32 --- 13.99 - 2.90 a 3.05 2.14 * 13 probability /0.05 *** *** probability /0.01 225 JACKSON 025666 TABLE 71 ODDS RATIOS FOR COMPANY A SUBJECTS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE 5 YEARS VS. NONE OTG BATCH PREPARATION CASES EEXXPPEXPEXP UNEXP 7 41 CONTROLS 16 162 VALUE OR 1.17 2.03 CONFIDENCE INTERVAL 2 MILLING 2 45 15 1.26 0.29 3 CALENDER 8 39 19 1.00 1.58 4 TUDER 7 41 9 157 1.62 2.26 5 FABRICATION 10 37 17 142 1.84 2.27 5 TUBES -FABRICATION 5 40 8 141 1.27 2.16 7 CURE PREPARATION 1 44 0.02 1.02 CURING 9 INSPECTION . 10 | MAINTENANCE 11 GENERAL SERVICE 3 38 11 132 1.41 45 9 152 0.19 37 20 124 1.04 10 27 33 81 0.11 1.94 1.13 0.53 0.94 12 SHIP & REC 13 RECLAIM 4 46 10 164 0.04 1.03 2 48 14 161 1.02 0.47 14 CHEMICALS 15 PLASTICS 48 6 178 0.53 0.50 1 52 197 == INSPECTION 2 16 PREP - OTHER 17 FINAL .46 1.77 47 3 169 0.91 6.2 0.00 18 METAL . HANDLING OW 25 28 73 0.09 0.96 19. OTHER RUBBER* Oo 51 191 - - 20 SALARY 21. UNKNOWN 50 3 43 178 0.20 0.80 0.27 1.21 * = probability /0.05 ** * probability /0.01 226 JACKSON 025667 TABLE 72 ODDS OTG BATCH . PREPARATION MILLING CALENDER TUBER RATIOS FOR COMPANY A SUBJECTS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE 10 YEARS VS. NONE CASES EXP UNEXP CONTROLS EXP UNEXP CHI VALUE OR CONFIDENCE LOW & 41 10 149 0.67 1.61 0.50 1 45 12 152 1.70 0.00 5 39 15 134 0.14 1.08 3 41 6 146 0.78 1.70 INTERVAL HIGH FABRICATION w 7 FABRICATION 40 CURE PREPARATION CURING WAWA 5 WAWA INSPECTION . 2 WAWA MAINTENANCE GENERAL SERVICE WAWA 2 27 & SHIP REC RECLAIM 46 2 48 CHEMICALS PLASTICS 48 0 52 13 134 1.31 5 135 1.07 1 155 0.93 9 125 0.86 8 148 0.26 12 121 0.77 15 75 0.10 6 161 0.14 9 161 0.45 5 178 0.38 i) 197 1.91 2.43 3.75 1.60 0.81 0.57 0.94 1.12 0.70 0.59 PREP - OTHER 46 0 168 0 FINAL INSPECTION METAL HANDLING 2 169 15 63 0 OTHER RUBBER 0 SALARY 0 191 2 174 UNKNOWN 1 43 144 227 = probability /0.05 *** *** probability /0.01 JACKSON 025668 TABLE 73 ODDS RATIOS FOR COMPANY B CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE 2 YEARS VS. NONE OTG 3 BATCH PREPARATION CASES EXP UNEXP 11 34 CONTROLS EXP = UNEXP 28 131 CHI VALUE OR 0.89 1.46 CONFIDENCE INTERVAL LOW HIGH 0.73 2.91 2 MILLING 1 50 16 175 1.57 0.24 0.05 1.07 3 CALENDER 5 38 38 104 2.38 * 0.26 0.10 0.66 4 TUBER 2 44 14 147 0.98 0.47 0.13 1.67 5 FABRICATION 12 29 6 TUBES FABRICATION 7 37 7 CURE PREPARATION 41 8 CURING 8 34 9 INSPECTION . 10 35 28 102 1.17 35 116 1.08 13 142 0.26 33 111 0.53 20 126 1.15 1.66 0.63 0.84 0.79 1.75 0.81 0.31 0.28 0.33 0.78 3.39 1.27 2.56 1.63 3.93 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 10 35 12 32 5 9 1 44 38 130 0.22 52 89 1.43 13 144 0.74 0 159 0.72 0.90 0.40 0.54 0.26 1.57 0.58 0.48 0.09 2.02 1.10 4.28 2.52 14 CHEMICALS 15 PLASTICS 0 97 3 45 5 183 1.15 14 156 0.43 0.00 --- --- 0.73 0.21 2.49 16 PREP = OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER . 20 SALAKY 21 UNKNOWN 11 31 S 40 25 113 1.29 15 139 0.20 1 49 7 182 0.83 4 43 23 146 0.90 ] 48 1 45 2 0.58 10 .158 1.03 1.71 0.86 3.39 1.12 0.45 2.76 0.48 0.11 2.07 0.60 0.24 1.52 2.00 0.28 14.43 0.36 0.07 1.85 * = probability 0.05 ** " probability 10.01 228 JACKSON 025669 TABLE 74 ODDS RATIOS FOR COMPANY B SUBJECTS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE 5 YEARS VS. NONE CASES CONTROLS CONFIDENCE INTERVAL OTC EXP EXP UNEXP CHI VALUE OR LOW BATCH PREPARATION 10 34 | 22 129 1.18 1.66 0.82 MILLING 49 CALENDER 38 TUBER 44 FABRICATION 29 37 TUBES FABRICATION CURE PREPARATION 41 34 CURING INSPECTION . 35 MAINTENANCE GENERAL SERVICE 10 35 11 32 SHIP & REC 40 RECLAIM 44 CHEMICALS PLASTICS PREP - OTHER FINAL INSPECTION METAL HANDLING OTHER RUBBER SALARY UNKNOWN 0 50 45 31 40 49 43 48 0 45 6 1.75 22 99 5 147 21 200 23 108 S 139 22 107 12 117 37 130 35 86 11 142 3 159 1 183 12 13 104 7 5 > fe 0.46 2.17 0.34 1.49 1.67 0.36 0.39 0.47 0.14 0.74 0.56 0.18 0.58 0.88 1.61 1.06 0.53 0.46 2.00 1.11 0.62 0.11 0.19 0.05 1.32 0.34 1.97 0.39 1.36 0.93 . 0.16 0.33 0.83 0.39 1.30 0.93 0.71 0.52 0.42 0.33 1.44 0.49 1.23 0.19 0.00 0.43 0.09 2.33 0.98 1.95 0.69 0.63 0.15 0.58 0.08 ow 0.00 1.90 3.24 2.06 1.53 4.17 7.89 2.03 5.52 5.50 2.64 4.07- 229 * 3probability 50.05 ** == probability 10.01 JACKSON 025670 TABLE 75 ODDS RATIOS FOR COMPANY B CASES COMPARED WITH THEIR MATCHED CONTROLS . EXPOSURE 10 YEARS VS. NONE OTG 1 BATCH PREPARATION 2 MILLING CASES EXP UNEXP 6 34 1 49 CONTROLS EXP UNEXP 16 118 4 175 . CHI VALUE OR 0.43 , 1.30 0.11 0.89 CONFIDENCE INTERVAL LOW HIGH 0.54 0.15 3 CALENDER 3 83955 19 99 1.95 0.22 0.06 4 TULER 1 83955 0 144 1.73 as --- --- 5 FABRICATION 9 83955 15 97 1.61 2.16 0.98 4.74 6 FABRICATION 1 7 CURE PREPARATION 1 8 CURING 2 83955 14 100 1.94 41 2 136 0.41 34 140 95 1.70 0.17 1.67 0.25 ; 0.04 0.21 0.07 0.76 12.95 0.95 9 INSPECTION . 4 35 8 115 0.46 1.33 0.47 3.73 10 MAINTENANCE 9 35 33 130 0.12 1.06 0.48 2.32 11 GENERAL SERVICE 7 32 22 83 0.34 0.84 0.36 1.97 12 SHIP & REC 13 RECLAIM 3 40 1 44 8 139 0.49 2 159 0.49 1.42 0.44 4.59 1.78 0.26 12.23 14 CHEMICALS 0 50 L 183 0.58 0.00 =< --- 15 PLASTICS 0 45 2 148 0.86 0.00 . oon cere 16 PREP - OTHER 5 31 8 97 1.21 2.21 " 0.75 6.50 17 FINAL INSPECTION 3 40 3 134 1.51 3.41 0.90 12.99 18 METAL HANDLING 1 49 3 182 0.12 . 0.89 0.19 4.23 19 OTHER RUBBER Q 43 1 133 0.50 0.00 o-- -- 20. SALARY 0 48 0 .186 -~ -- " - ~ Lt) I 230 21 UNKNOWN -- 45 155 0.50 0.00 --- * * probability /0.05 ** 16 probability /0.01 e JACKSON 025671 TABLE 76 ODDS RATIOS FOR ALL CONFIRMED CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES CONTROLS CONFIDENCE INTERVAL : 2 BATCH PREPARATION .17 47 54 195 0.89 1.41 0.75 2.65 2. MILLING 5 31 218 1.17 0.53 0.21 1.30 3 CALENDER 16 69 180 0.50 0.85 0.49 2.46 4 TUBER 10 34 215 0.46 1.19 0.64 2.19 5 FABRICATION 22 72 177 0.83 1.31 0.77 2.22 6. FABRICATION 15 69 180 0.72 0.78 0.44 1.38 7 CURE PREPARATION 9 31 218 0.41 1.18 0.61 2.27 8 CURING 15 71 178 0.86 0.76 0.44 1.29 9 INSPECTION 12 73 176 1.10 0.65 0.34 1.24 10 MAINTENANCE 18 55 194 1.08 1.44 0.82 2.51 11 GENERAL SERVICE 27 115 134 0.58 0.84 0.51 1.38 12. SHIP & REC 11 35 214 0.64 1.29 0.67 2.45 13 RECLAIM 5 28 221 0.81 0.67 0.29 1.52 3 11 238 0.07 1.05 0.33 3.36 14 CHEMICALS 15 PLASTICS 2 18 231 1.22 0.36 0.09 1.43 16 PREF ~ OTHER 12 8 17 FINAL INSPECTION 44 205 0.16 35 214 0.35 1.06 0.60 1.86 0.86 0.41 1,77 | 21 13 METAL HANDLING 61 188 1.50 1.72 0.95 3.12 32 217 1.50 0.44 0.18 1.08 19 OTHER RUBBER , 11 238 0.68 1.50 0.56 4.00 20. SALARY 35 214 0.02 1.01 0.46 2.20 210 UNKNOWN 231 231 * = probability 0.05 ** = probability /0.01 JACKSON 025672 TABLE 77 ODDS RATIOS FOR ALL UNCONFIRMED CASES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES ORG UNEXP CHI VALUE OR CONFIDENCE INTERVAL LOW HIGH BATCH PREPARATION 28 27 127 1.21 1.59 0.85 2.99 MILLING 35 18 136 0.91 0.50 0.14 1.73 CALENDER 30 45 109 0.95 0.69 0.37 1.31 TUBER 31 25 129 0.06 1.03 0.49 2.14 FABRICATION 14 24 47 107 0.55 1.26 0.63 2.54 TUBES FABRICATION 10 - 28 34 120 0.42 1.17 0.63 2.19 CURE PREPARATION 30 18 136 1.44 1.88 0.92 3.88 CURING 15 23 32 122 2.26 2.29 1.26 4.19 INSPECTION 10 28 22222 127 1.20 1.84 0.79 4.25 MAINTENANCE 10 28 22222 116 0.07 1.03 0.49 2.15 GENERAL SERVICE 16 22 22222 102 0.49 1.19 0.67 2.10 SHIP & REC www 22222 131 0.38 0.83 0.37 1.85 RECLAIM www 22222 134 0.10 0.95 0.41 2.21 CHEMICALS www 9 145 0.83 0.44 0.09 2.21 PLASTICS PREP - OTHER 35 18 136 0.83 12 26 2237 129 2.05 0.54 0.16 1.83 2.49 1.20 5.17 FINAL INSPECTION 7 31 2237 133 0.66 1.40 0.60 3.27 METAL HANDLING OTHER RUBBER 31 2237 120 1.17 4 34 15 139 0.10 ~ 0.53 0.22 1.29 1.08 0.30 3.84 , 38 7 . 147 1.32 0.00 no 232 SALARY UNKNOWN 32 26 128 0.24 0.88 0.38 - 2.06 * 19 probability /0.05 ** "probability 10.01 JACKSON 025673 TABLE 78 OTG 1 BATCH PREPARATION 2 MILLING ODDS RATIOS FOR ALL CASES WHO ENTERED THE INDUSTRY AT AGE 24 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES CONTROLS EXP UNEXP CHI VALUE OR CONFIDENCE LOW 12 28 40 115 0.58 1.28 0.64 3 37 22 133 1.10 0.49 0.17 1.43 3. 15 25 46 109 0.71 1.29 0.72 2.32 4 9 31 28 11227 7 0.65 1.29 0.68 2.47 24 16 60 95 2.41 * 2.47 1.33 .4.57 10 30 42 113 0.24 0.89 0.41 1.95 CURE PREPARATION 12 28 18 2.90 * 2.93 1.59 5.40 CURING 18 22 47 1.82 1.80 1.06 3.06 INSPECTION MAINTENANCE 14 26 44 14 26 38 0.82 . 1.37 1.48 0.68 3.25 1.72 0.90 3.29 GENERAL SERVICE SHIP & REC 17 23 70 7 33 27 0.31 0.90 0.51 0.02 0.99 0.45 1.60 RECLAIM 35 16 0.44 1.25 0.55 2.84 CHEMICALS 37 0.38 1.31 0.40 4.30 PLASTICS ) 40 14 2.11 * 0.00 --- PREP - OTHER 13 27 30 1.90 2.16 1.11 FINAL INSPECTION 6 34 23 0.03 0.99 , 0.40 METAL HANDLING OTHER RUBBER , SALARY UNKNOWN 8 32 38 0.67 3 37 19 136 0.90 2 38 9 146 0.16 ? 33 25 25 25 130 0.21 0.70 0.30 0.51 0.15 0.88 0.23 1.10 0.51 * A probability /0.05 ** - probability 50.01 233 JACKSON 025674 TABLE 79 ODDS RATIOS FOR ALL CASES WIND ENTERED THE INDUSTRY AT AGE 24 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1 BATCH PREPARATION 2 MILLING CASES EXP UNEXP 15 47 -- --,--,, 57 CONTROLS EXP = UNEXP 41 197 27 211 CHI VALUE OR 1.40 1.67 0.99 0.54 CONFIDENCE INTERVAL - LOW HIGH 0.92 3.03 0.20 1.49 3 CALENDER -- --,--,, 52 65 ' 173 1.88 0.50 0.28 0.92 4 TUBER 8 54 31 207 0.09 0.96 0.48 1.92 5 FABRICATION 12 50 59 179 0.89 0.72 0.39 1.33 6 TUBES -FABRICATION 15 7 CURE PREPARATION 5 S. CURING 12 9 INSPECTION 8 47 61 57 31 50 57 54 46 177 0.24 207 1.02 181 0.67 192 1.12 - 0.93 0.57 1.53 0.58 0.25 1.39 0.79 0.44 1.42 0.63 0.32 1.24 - 10 MAINTENANCE 11 GENERAL SERVICE 14 . 26 48 55 36 98 183 0.00 140 0.02 oo 1.00 --- 1.01 0.62 | 1.65 12 SHIP & REC 9 43 31 207 0.38 1.17 0.59 2.32 13. RECLAIM 5 57 32 206 1.15 0.56 0.24 1.28 14 CHEMICALS 15 PLASTICS . 1 61 11 227 1.06 S 57 22 216 0.25 0.36 0.08 1.76 0.85 0.30 .2.44 16 PREP - OTHER 11 51 39 199 0.27 1.10 0.60 2.01 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20 SALARY 9 _ 20 5 2 53 33 42 61 57 28 60 9 5 36 205 0.16 1.13 210 0.79 229 0.15 202 0.39 1.07 " 0.53 1.50 0.83 0.65 0.27 0.89 0.25 0.83 0.37 2.18 2.71 1.59 3.20 1.86 234 21 UNKNOWN it E probability /0.05 ** = probability /0.01 JACKSON 025675 TABLE 80 ODDS RATIOS FOR COMPANY A SUBJECTS WHO ENTERED THE INDUSTRY AT ACE COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER /24 OTG 1 BATCH PREPARATION 2 MILLING CASES EXP UNEXP 4 14 2 16 CONTROLS EXP UNEXP 11 59 93 60 CHI VALUE OR 0.81 1.83 0.31 0.75 CONFIDENCE LOW 0.54 0.16 3 CALENDER 8 10 11 59 2.77 4.76 1.89 4 TUBER 6 12 10 60 1.86 2.49 1.11 5 FABRICATION 9 6 -FABRICATION 4 9 23 47 1.28 14 14 56 0.26 1.91 0.83 1.19 0.39 7 CURE PREPARATION 8 CURING 9 INSPECTION 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14. CHEMICALS . 15 PLASTICS 16 PREP - OTHER 17 FINAL INSPECTION 18 METAL HANDLING | 19. OTHER RUBBER 20 SALARY 21. UNKNOWN 4 14 6 33333 1.77 3 10 17 33333 1.73 2 16 14 56 0.91 6 12 20 33333 0.43 10 26 44 1.40 3 15 1 17 3 15 9 61 8 26 26 0.44 0.76 1.89 0 18 9 70 _ 2 16 0. 70 1.19 2 16 A 66 0.74 8 8 33 37 0.21 1 17 1 69 0.93 4 17 9 61 0.86 4 14 10 60 0.81 3.00 2.34 0.38 1.28 2.00 1.44 0.50 5.25 and 2.13 2.17 0.89 3.75 0.38 1.64 1.08 1.04 0.07 0.49 0.89 0.37 0.11 1.24 -- 0,75 0.39 0.36 0.36 0.06 0.60 2.19 3.34 4.50 5.59 2.26 22.25 ae 6.06 12.04 2.18 38.56 2.46 4.48 ^ probability ** /0.05 probability /0.01 235 JACKSON 025676 TABLE 81 ODDS RATIOS FOR COMPANY A SUBJECTS WHO ENTERED THE INDUSTRY AT AGE COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER 24 CASES OTG EXP CHI VALUE CR CONFIDENCE INTERVAL LOW HIGH BATCH PREPARATION MILLING .S CALENDER S 27 17 2 18 2 26 110 1.22 109 0.10 101 0.73 1.96 0.93 0.69 0.79 0.31 0.29 4.83 2.82 1.60 TUBER 2 14 113 0.47 1.31 0.51 3.41 8 FABRICATION TUBES -FABRICATION CURE PREPARATION CURING 6 INSPECTION 5 10 MAINTENANCE 11 GENERAL SERVICE 15 15 12 SHIP & REC 3 13 RECLAIM 3 14 CHEMICALS I 15 PLASTICS oO 16 PREP - OTHER Vv 17 FINAL INSPECTION 3 18 METAL HANDLING 19 19 OTHER RUBBER 20 SALARY 211 UNKNOWN 2 19 108 0.32 2 22 105 0.80 30 229287 28 229287 29 229287 27 229287 19 229287 31 15 115 0.51 100 0.40 108 0.08 104 0.43 77 0.43 112 0.44 31 21 33 4 34 0 31 7 31 8 106 1.10 123 0.00 127 120 0.61 119 0.45 15 52 75 1.71 34 2 125 0.71 33 7 120 0.53 28 23 _ 104 0.06 1.20 0.47 3.02 1.46 0.68 3.13 1.42 0.46 4.39 0.81 0.34 1.94 1.04 0.45 2.43 1.22 0.57 2.64 1.20 0.59 2.43 0.75 0.26 2.17 0.50 0.15 1.41 1.00 1.54 1.37 2.01 0.00 0.57 1.04 * probability 0.05 *** *** probability /0.01 236 JACKSON 025677 TABLE 82 > ODDS RATIOS FOR COMPANY B SUBJECTS WHO ENTERED THE INDUSTRY AT AGE 24 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES CONFIDENCE INTERVAL OTG UNEXP EXP UNEXP CHI VALUE OR LOW BATCH PREPARATION 8 14 29 56 0.20 1.10 0.48 2 MILLING 1 21 12 73 1.21 0.32 0.07 1.51 3 CALENDER TUBER 7 15 39 47 1.16 3 19 18 67 0.79 0.54 0.22 1.29 1.29 1.S 0.59 0.19 1.S1 1.S1 FABRICATION 6 TUBES FABRICATION 7 CURE PREPARATION 7 37 16 28 14 12 48 2.08 * 3.23 1.28 8.17 8.17 0.48 2.31 * 0.73 2.90 0.24 1.36 2.16 2.16 2.16 6.20 6.20 8 CURING 8 INSPECTION 12 30 10 27 0.92 1.69 1.49 0.73 3.02 2.61 1.03 6.64 10 MAINTENANCE 11 GENERAL SERVICE 8 14 18 7 15 44 1.53 2.30 0.94 0.97 1.70 0.39 0.16 0.97 0.97 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP - OTHER & 18 18 0Ff 18 8 22 C 22 14 11 11 27- 0.34 0.82 0.33 5.81 1.16 2.07 0.74 1.22 2.11 * 0.00 0.00 1.58 2.17 17 FINAL INSPECTION 18 19 0.44 0.75 18 METAL HANDLING Oo 22 5 1.18 0.00 19 OTHER RUBBER 20 SALARY 21 UNKNOWN 4 20 18 1.32 0.28 21 0 2.00 * weere 19 15 70 0.45 0.72 237 * probability /0.05 ** = probability 0.01 JACKSON 025678 TABLE 83 ODDS RATIOS FOR COMPANY B SUBJECTS WHO ENTERED THE INDUSTRY AT ACE 24 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1. BATCH PREPARATION CASES EXP = UNEXP 8 20 2. MILLING 3 CALENDER 0 28 5 23 4 TUBER 3 22222 5 FABRICATION 6 22222 6 TUBES FABRICATION ? 22222 7 CURE PREPARATION 1 22222 8 CURING 6 22222 9 INSPECTION 10 MAINTENANCE 3 2222222222 7 2222222222 11 GENERAL SERVICE 11 22222 12 SHIP & REC 6 22222 13 RECLAIM 12 22222 14 CHEMICALS 0 22222222 15 PLASTICS 5 22222222 10 PREP - OTHER 17 FINAL INSPECTION INSPECTION 18 METAL HANDLING 8 22222222 6 22222222 1 22222222 19 OTHER RUBBER S 22222222 20 SALARY 1 22222222 21. .- UNKNOWN 2 22222222 CONTROLS CONTROLS EXP UNEXP 24 87 9 102 39 72 17 94 40 71 31 80 19 92 29 82 27 84 32 79 48 63 16 95 11 100 7 104 22 .89 32 79 21 90 9227 101 9227 85 _ 9227 109 9227 98 CHI VALUE OR 0.79 1.47 CONFIDENCE INTERVAL Haale 3.28 1.59 1.95 0.61 0.00 0.36 0.69 --~ 0.15 0.25 --- 0.85 1.90 1.42 1.10 1.85 0.48 0.20 1.12 0.64 0.32 1.25 0.17 0.03 0:82 0.55 0.77 0.34 1.70 1.66 0.32 0.10 0.99 0.42 0.78 0.29 2.03 0.44 0.83 0.41 1.67 0.92 1.67 0.67 4.17 0.46 0.68 0.17 2.70 1.37 0.25 0.05 0.00 0.85 0.99 --- 0.30 0.48 --- 2.44 2.02 0.14 1.01 0.62 0.58 0.67 0.93 0.41 0.71 2.00 0.58 0.38 0.10 0.29 0.28 0.15 2.28 1.76 1.75 14.43 2.20 238 * = probability /0.05 ** = probability /0.01 JACKSON 025679 TABLE 84 ODDS RATIOS FOR A COMPANY SUBJECTS WHO DIED AT AGE /65 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTG 1 BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION 6 FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION . 10 MAINTENANCE GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP + OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20 SALARY 21 UNKNOWN CASES EXP UNEXP EXP 3 9 7 2 10 8 6 6 11 9 10 5 7 18 4 & 12 3 9 6 6 6 10 z 11 10 5 7 13 9 3 14 3 9 5 1 11 7 3 9 4 0 12 f) 2 10 L 2 10 L 7 5 25 1 11 0 1 11 9 2 10 5 CONFIDENCE INTERVAL CHI VALUE OR LOW HIGH 39 1.02 2.25 0.61 8.31 38 0.00 1.00 35 1.80 36 0.27 28 0.13 34 0.60 40 1.10 36 2.01 36 1.06 3.24 1.11 1.11 9.49 1.19 0.41 0.41 3.44 9.4 1.09 0.36 3.26 1.52 0.49 4.734.73 4.73 2.00 0.71 5.67 3.29 1.24 8.72 0.24 0.02 2.21 33 0.58 1.62 0.57 4.60 4.60 4.60 32 2.55 5.76 1.86 1.86 1.86 41 1.07 2.35 39 0.61 _ 0.57 43 1.89 5.25 46 we 45 0.80 1.68 45 2.03 --- 21 0.27 46 1.73 1.19 Smee? 0.86 0.38 , 41 0.53 1.55 6.01 * m probability /0.05 *** = probability /0.01 239 JACKSON 025680 TABLE 85 ODDS RATIOS FOR COMPANY A SUBJECTS WHO DIED AT ACE 65 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES OTG EXP EXP UNEXP EXP CHI VALUE OR BATCH PREPARATION o| 32 21 130 1.12 1.81 0.76 MILLING 35 20 131 0.33 0.80 0.28 CALENDER TUBER 8 33 26 32 14 FABRICATION 10 30 24 TUBES FABRICATION 8 32 24 125 0.03 137 1.76 127 1.29 127 0.58 1.01 0.48 2.27 1.06 1.80 0.85 3.79 1.31 0.61 2.79 CURE PREPARATION CURING INSPECTION . 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 35 12 139 1.03 5 32 34 117 0.27 23 23 128 0.08 an 23 29 122 0.21 16 24 61 90 0.12 27 18 133 0.76 3 37 22 129 1.18 1.97 0.67 0.83 0.42 1.8 1.04 0.45 2.38 1.10 0.52 2.30 0.95 0.50 0.60 0.20 1.32 0.48 0.17 1.34 14 CHEMICALS 15 PLASTICS 39 4 147 0.00 1.00 0 40 151 3 16 PREP - OTHER 17 FINAL INSPECTION 18. METAL HANDLING .20 19 OTHER RUBBER a 20 SALARY 21 UNKNOWN 37 37 11 20 - 60 40 3 32 7 32 28 145 0.82 140 0.01 .91 1.25 148 0.87 144 0.53 123 0.34 1.78 1.01 1.59 0.00 0.57 1.19 K = probability 10.05 ** * probability /0.01 240 JACKSON 025681 TABLE 86 ODDS RATIOS FOR COMPANY B CASES WHO DIED AT AGE 65 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER OTC 1 BATCH PREPARATION 2 MILLING 3 CALENDER 4 TUBER 5 FABRICATION 6 TUBES -FABRICATION 7 CURE PREPARATION 8 CURING 9 INSPECTION . - 10 MAINTENANCE 11 GENERAL SERVICE 12 SHIP & REC 13 RECLAIM 14 CHEMICALS 15 PLASTICS 16 PREP M OTHER 17 FINAL INSPECTION 18 METAL HANDLING 19 OTHER RUBBER 20. SALARY 21 UNKNOWN CASES EXP UNEXP 3 8 1 10 1 10 1 29 2 29 2 9 0 11 S 6 2 9 3 8 4 7 2 9 2 9 0 11 2 9 5 6 4 7 0 11 2 9 0 11 1 10 CONTROLS EXP UNEXP 10 32 4 38 16 26 8 34 10 32 12 30 2 40 13 29 6 36 14 28 CHI VALUE OR 0.20 1.15 0.00 1.00 2.10 0.00 0.78 0.43 1.03 0.36 1.08 0.38 0.82 0.00 0.81 1.71 0.30 1.43 0.36 0.69 CONFIDENCE INTERVAL LOW 0.35 3.80 --- --- on --- 0.07 0.07 0.09 -~- 2.53 1.84 1.66 --- 0.58 5.03 0.20 10.25 0.13 3.73 18 24 0.48 6 36 0.35 3 39 1.10 5 37 1.19 13 29 0.82 0.71 1.41 3.11 0.00 0.48 10.22 0.28 0.57 --- 0.11 2.31 7.08 17.01 --- 2.10 17 25 0.26 12 233 0.47 7 233 1.70 14 233 0.00 0 23 - 4 38 0.00 1.17 1.45 0.00 1.00 one 1.00 0.43 0.40 ----onn ana 3.23 5.23 --~ ----- 241 * = probability /0.05 ** m probability 0.01 JACKSON 025682 TABLE 87 ODDS RATIOS FOR COMPANY B CASES WHO DIED AT ACE 65 COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER CASES CONTROLS CONFIDENCE INTERVAL OTG EXP UNEXP EXP UNEXP CHI VALUE OR LOW RIGH EATCH PREPARATION 13 26 43 111 0.66 1.31 0.67 2.55 2.55 2.55 MILLING 0 39 17 137 2.20 0.00 CALENDER TUEER 11 28 61 93 1.44 5 34 27 127 0.71 0.57 0.69 0.30 0.30 1.61 1.61 1.61 FABRICATION 19 20 TUBES -FABRICATION 11 28 CURE PREPARATION 9 30 CURING 11 28 63 91 0.87 53 1.01 0.75 29 125 0.57 46 103 0.19 1.40 0.74 2.67 0.75 1.41 1.41 .41 0:51 1.25 2.37 0.93 1.71 - INSPECTION 13 26 51 103 0.02 0.99 0.47 0.47 2.08 MAINTENANCE - 12 27 36 118 0.94 GENERAL SERVICE - 14 25 74 80 1.41 SHIP & REC RECLAIM 8 31 28 126 0.29 4 35 16 138 0.03 0.74 1.09 1.50 3.04 0.59 1.13 2.28 2.8 2.8 0.98 0.39 2.46 CHEMICALS 39 8 146 1.41 0.00 PLASTICS 36 23 131 1.22 0.42 0.13 1.36 , PREP - OTHER 14 25 42 112 1.04 1.48 0.80 2.73 FINAL INSPECTION 6 33 32 122 0.79 0.67 0.30 1.53 METAL HANDLING | OTHER RUBBER , 20 SALARY UNKNOWN 1 38 .7 147 0.55 5 34 36 118 1.41 2 37 2 152 1.50 4 35 24 130 0.86 0.57 0.11 3.07 0.50 0.22 1.12 4.00 0.87 18.29 0.60 0.23 1.59 * = probability /0.05 ** = probability /0.01 242 JACKSON 025683 TABLE 88 EXPOSURE = NEVER a 1. High Exposure 2. High or Mod Exp ODDS RATIOS FOR COMPANY A SUBJECTS WITH POTENTIAL EXPOSURE TO POLYCYCLIC HYDROCARBONS COMPARED WITH THEIR MATCHED CONTROLS CASES EXP UNEXP CONTROLS CONFIDENCE INTERVAL EXP UNEXP CHI VALUE OR LOW HIGH 17 35 27 25 54 143 0.87 97 100 0.34 1.34 1.10 0.77 0.68 2.35 1.79 EXPOSURE 22 yrs vs. N. 2. High or Mod Exp 20 23 63 78 0.21 1.08 0.61 1.91 EXPOSURE 5yrs vs. N. 1. High Exposure 2. High or Mod Exp 8 34 17 23 23 109 0.05 48 71 0.14 EXPOSURE 210 yrs vs N. 1. High Exposure 6 33 14 107 0.60 2. High or Mod Exp 11 23 30 68 0.44 * = probability 0.05 ** = probability 0.01 a See Appendix III and Table 98 0.98 1.05 0.46 0.56 2.09 1.97 1.37 0.58 3.25 1.19 0.62 2.31 243 JACKSON 025684 TABLE 89 ODDS RATIOS FOR COMPANY B SUBJECTS WITH POTENTIAL EXPOSURE TO POLYCYCLIC HYDROCARBONS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER 1. High Exposure 2. High or Mod Exp CASES EXP UNEXP CONTROLS ; CONFIDENCE INTERVAL EXP UNEXP CHI VALUE OR LOW HIGH 22 28 92 104 | 0.40 | 22 19 123 73 0.10 0.88 0.52 0.96 0.53 1.50 1.76 EXPOSURE 2yrs vs. N. 1. High Exposure 17 27 57 88 | 0.14 2. High or Mod Exp . 25 18 81 63 0.18 EXPOSURE 25 yrs vs. N. 1. High Exposure 15 27 38 85 0.69 2. High or Mod Exp 24 18 63 62 0.67 0.95 1.07 0.51 0.56 1.78 2.04 1.32 1.30 0.68 0.68 2.58 2.49 EXPOSURE 10 yrs vs. N. 1 High Exposure 10 27 25 2. High or Mod Exp _ 18 18 40 75 | 0.33 0.36 | 1 .17 0.53 2.57 1.17 0.58 2.35 pr obability 0.05 ** = probability 0.01 244 JACKSON 025685 TABLE 90 ODDS RATIOS FOR COMPANY A SUBJECTS WITH POTENTIAL EXPOSURE TO NITROSAMINES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER a 1. High Exposure 2. High or Mod Exp CASES EXP UNEXP CONTROLS . CONFIDENCE INTERVAL EXP UNEXP = CHI = VALUE = LOW HIGH 18 34 54 143 | 0.95 _ 31 21 104 93 | 0.91 0.79 1.30 0.81 2.35 2.10 EXPOSURE 2yrs va N. 1. High Exposure 10 32 {| 2. High or Mod Exp 27 19 36 110 {| 0.15 | 72 78 1.53 EXPOSURE 5yrs vs. N. 1. High Exposure 2. High or Mod Exp 6 32 21 19 19 102 | 0.20 5 61 1.08 0.92 0.39 1.64 0.96 2.17 2.79 3 ' | 0.88 0.31 2.50 | 1.49 0.81 2.74 EXPOSURE 10yrs vs N. 1. High Exposure 2 32 12 2. High or Mod Exp 14 19 .34 | 93 | 1.56 | 51 | 0.40 0.14 0.02 1.11 1.19 | 0.59 2.42 * = probability 0.05 ** = probability 0.01 a See Appendix III and Table 98 245 JACKSON 025686 TABLE 91 ODDS RATIOS FOR COMPANY B CASES WITH POTENTIAL EXPOSURE TO NITROSAMINES COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER : 1. High Exposure 2. High or Mod Exp CASES EXP UNEXP CONTROLS EXP UNEXP = CHI = VALUE CONFIDENCE INTERVAL OR LOW HIGH 31 19 98 98 1.51 36 14 131 65 0.70 1.63 1.31 0.96 0.69 2.79 2.46 EXPOSURE 2 yrs vs. N. 1. High Exposure 23 19 | 2. High or Mod Exp 30 14 60 80 | 1.46 91 57 0.75 1.68 - 0.94 1.35 0.70 3.02 2.60 ' EXPOSURE 5yrs vs. N. 1. High Exposure High or Mod Exp 21 19 28 14 46 76 | 2.39 * 1.74 0.97 3.12 74 55 0.86 : 1.40 0.74 2.67 EXPOSURE 210 yrs vs. N. 1. High Exposure 13 19 2. High or Mod Exp | 19 14 28 61 0.69 48 45 0.34 1.33 0.67 2.64 1,16 0.57 2.33 * = probability 0.05 *** *** probability 0.01 246 JACKSON 025687 TABLE 92 EXPOSURE = NEVER 1. High Exposure 2. High or Mod Exp ODDS RATIOS FOR COMPANY ^ CASES WITH POTENTIAL EXPOSURE TO CARBON BLACK COMPARED WITH THEIR MATCHED CONTROLS CASES EXP UNEXP g 43 . 21 31 CONTROLS . EXP UNEXP CHI 21 |. 176 | 1.66 64 133 1.14 VALUE CONFIDENCE INTERVAL OR LOW HIGH 1. 2.31 1.011 5.29 1.49 0.84 2.66 EXPOSURE i 2yrs ys N. 1. High Exposure 2. High or Mod Exp 6 42 13 30 13 164 1.39 38 111 0.60 2.52 1.29 0.85 0.64 7.49 2.61 EXPOSURE 5 yrs vs. N. 1. High Exposure 2. High or Mod Exp 6 42 13 . 30 12 164 1.42 29 111 1.38 EXPOSURE 10yrs vs. N. 1. High Exposure 2. High or Mod Exp 4 42 7 159 | 1.36 * 7 29 18 97 0.42 | 2.56 0.86 7.59 1.87 0.89 3.93 | 2.94 0.80 10.87 1.28 0.48 3.43 * = probability 0.05 ** probability /0.01 a See Appendix III and Table 98 247 JACKSON 025688 TABLE 93 ODDS RATIOS FOR COMPANY B SUBJECTS WITH POTENTIAL EXPOSURE TO CARBON BLACK COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE NEVER CASES - EXP UNEXP CONTROLS EXP UNEXP , CHI = VALUE 1. High Exposure 14 36 41 155 1,06 2. High or Mod Exp ss 33 17 103 93 1.70 CONFIDENCE INTERVAL OR LOW HIGH 1.45 . ! ! 0.81 2.59 1.74 1.02 2.97 EXPOSURE 2yrs vs. N. 1. High Exposure 2. High or Mod Exp 9 36 24 17 28 138 | 1.02 28 69 | 1.23 1.58 0.75 3.31 1.58 0.86 2.90 EXPOSURE > yrs vs. N. 1. High Exposure 2. High or Mod Exp 8 36 22 17 16 135 | 1.23 46 63 | 1.26 EXPOSURE 10 yrs vs N. 1. High Exposure S 36 2. High or Mod Exp 14 17 10 126 | 0.99 30 52 | 0.90 1.79 1.66 0.82 0.86 3.88 3.23 it 1.76 0.69 4.50 1.56 0.69 2.51 248 * = probability /0.05 ** = probability 0.01 JACKSON 025689 TABLE 94 ODDS RATIOS FOR COMPANY A CASES WITH POTENTIAL EXPOSURE TO ASBESTOS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER a 1. High Exposure 2. High or Mod Exp CASES EXP UNEXP CONTROLS EXP UNEXP , CHI = = VALUE = OR CONFIDENCE INTERVAL LOW HIGH 13 39 40 157 0.72 | 1.32 0.70 2.50 27 25 64 133 2.69 ** 2.41 1.41 4.13 EXPOSURE 2yrs vs. N. 1. High Exposure 2. High or Mod Exp 9 39 16 25 19 145 | 1.14 28 108 2.26 i 1.70 0.79 3.68 * 2.48 1.28 4.81 EXPOSURE 25 yrs vs. N. 1. High Exposure 2. High or Mod Exp 7 39 13 25 9 140 | 1.86 18 101 __| 2.37 2.70 1.12 * 2.75 1.36 6.50 5.56 EXPOSURE 10 yrs vs N. 1. High Exposure 2. High or Mod Exp 3 39 4 7 | 25 9 129 | 1.33 87 1.85 * Z probability 0.05 *** *** probability 0.01 a See Appendix III and Table 98 3:18 0.76 2.72 | 1.12 13.25 6.65 249 JACKSON 025690 TABLE 95 ODDS RATIOS FOR COMPANY B SUBJECTS WITH POTENTIAL EXPOSURE TO ASBESTOS COMPARED WITH THEIR MATCHED CONTROLS EXPOSURE = NEVER 1. High Exposure 2. High or Mod Exp CASES EXP UNEXP CONTROLS EXP CONTRLS UNEXP = CHI VALUE = OR CONFIDENCE INTERVAL LOW HIGH 12 38 78 118 2.14 33 17 122 74 0.46 * 0.46 0.26 1.16 0.68 0.84 2.00 EXPOSURE 2yrs vs. N. 1. High Exposure 2. High or Mod Exp 9 37 26 16 46 105 1.46 76 58 0.64 0.56 0.29 1.27 0.68 1.08 2.35 EXPOSURE 25 yrs vs. N. 1. High Exposure 2. High or Mod Exp 4 37 32 21 16 49 94 | 2.20 51 0.82 * 0.29 0.11 -| 0.73 1.38 0.72 2.64 EXPOSURE 10yrs vs N. 1. High Exposure 2 37 20 89 2.20 * 0.20 0.06 0.66 2. High or Mod Exp 14 16 30 43 0.52 1.26 0.60 2.64 * = probability 0.05 ** = probability /0.01 250 JACKSON 025691 251 TABLE 96 UNMATCHED ODDS RATIOS FOR POTENTIAL ASBESTOS EXPOSURE BY DECADE OF ENTRY INTO THE INDUSTRY DECADE 1910-1919 CASES EXP UNEXP 4 7 CONTROLS EXP UNEXP 5 18 CHI 0.81 2.05 1920-1929 1930-1939 12 9 25 55 0 2 9 13 4.80 1.31 2.93 1940-1949 6 11 19 51 0.44 1.46 1950-1959 1 0 2 2 0.83 JACKSON 025692 TABLE 97 UNMATCHED ODDS RATIOS FOR POTENTIAL ASBESTOS EXPOSURE AT VARYING LATENCY PERIODS CASES MEAN EXPOSURE AMONG EXPOSED EXP UNEXP IN MONTHS 13 15 88.1 5 5 81.8 5 9 17.8 CONTROLS MEAN EXPOSURE AMONG EXPOSED EXP = UNEXP IN MONTHS <fl OR Latency 40 46 119 60.2 30 < Latency 40 6 6 112.8 20 < Latency 30 6 14 61.3 2.24 0.14 1.29 Latency Latency defined as Year of Death - Year of Hire 252 JACKSON 025693 253 TABLE 98 COMPOSITION OF CHEMICAL - SPECIFIC OTGs II POTENTIAL ASBESTOS EXPOSURE 1 High Exposure - CMTR CRTU FITU FUSP MIMG TUBU 2 Moderate Exposure - BPMC BPTR BPTU INDA MITU MNMS MNCP SBPU SBPR STOX STTU 3 Low Exposure -- CEMG CRBA CRFB CRSX CRTR CRVV MIFB MIMS MIPL MITD MNMA MNWR OTHR PALI PIBE PWPT RCMI SREC TCKR TEST TUBC VVMA ' 4 No Exposure - All other OTs not included above POTENTIAL CARBON BLACK EXPOSURE 1 High Exposure BLAN BPMC BPTR BPTU PIBE SBPR SBPU 2 Moderate Exposure - CMTR INDA MNMS SJAN 3 Low Exposure - CRBA CRFB CRSX CRTR CRTU CRVV FIRA FIRB FITU FUSP INDC MIFR MIMG MIMS MIPL MITD MITU MNEC MNEL MNWR RDCV RCMI RCPR REDE SREC STOR TCKR TEST 4. No Exposure - All others III POTENTIAL POLYCYCLIC HYDROCARBONS EXPOSURE High Exposure - BPMC BPTR BPTU CMTR CRBA CRFB CRSX CRTR CRTU CRVV MIMG PIBE RCDV Moderate Exposure - FIRB INDA MIFB MIMS MIPL MITD MITU RCMI RCPA SBPR SBPU SREC Low Exposure - BEBL CALO CALT CEMD CEMG FIRA FITU FUSP INDC INRE MNMS OTHR OTNT PLHA PWPT REDE SJAN SLIN SMOL SRLC STBB STOD STOR STOX STTU TCKR TEST TRBL TUBC TUBD TUBU VVMA 4. . No Exposure ~- All others IV POTENTIAL NITROSAMINES EXPOSURE 1. High Exposure - BPMC BPTR BPTU CRSX FOAM INDA MIFB MIPL MIMG MIMS MIPL MITD MITU 2. Moderate Exposure - CALO CALT CMTR CRBA CRFB CRTR CRTU CRVV FIRB RCDV RCMI SBPR SBPU SMOL TUBC TUBD TUBU . Low Exposure - BEBL CEMD SMOL OTHR ROCP RSMF - STOX STTU No Exposure - All others FIRA OTNT SJAN SYNT FITU PLHA SLIN TCKR FUSP PLST SREC TEST INDC RCPR SRLC TRBL MNEL REDE STOD URET MNWL ROCK STOR JACKSON 025694 TABLE 99 LATENCY,, SUMMARY OF FINDINGS FOR ASBESTOS EXPOSURE BY POTENTIAL LATENCY,, AND DECADE OF ENTRY POTENTIAL LATENCY MEAN POTENTIAL ASBESTOS EXPOSURE MONTHS % EXPOSED OR CHI 1 40 years cases controls 20 30-39 years cases controls 3 20-29 years cases controls i 41 13/28 = 46 3.88 2.24 3.88 18 46/165 = 28 5/10 ==== 41 50 _ 1.01.0 56 6/12 50 6 6/14 = 35 i 1.29 0.14 | 18 6/20 === 30 Entry Decade 1 1930 cases controls 2.61 41 16/32 * 50 , 5.21 18 28/101 * 28 2 1930 cases controls 1717 13/20 13/20 - 35 1.18 0.110.1 22 30/96 = 31 Latency defined as Year Out - Year In 254 JACKSON 025695 TABLE 99 cont'd LATENCY,, SUMMARY OF FINDINGS FOR ASBESTOS EXPOSURE BY POTENTIAL AND DECADE OF ENTRY POTENTIAL LATENCY | % FOREIGN BORN MEAN TOTAL | ACTIVE DURATION MEAN AGE AT HIRE MEAN AGE AT DEATH 1 40 years cases 22 controls 22 2 30-39 years cases ' 23 controls 33 3 20-29 years cases 22 controls 22 Entry Decade 1 1930 cases controls 2 1930 cases controls 32 _ 25 23 20 409 350 326 242 204 171 399 386 240 260 ; 23 73 - 31 28 63 ; I 40 45 47 ; 26 28 : 23 39 70 - \ 1 72 1 72 os Latency 255 defined as Year Out - Year In , BATO JACKSON 025696 DISCUSSION This investigation explored the possible related excess of stomach cancer while controlling for as many occupational confounders . as possible and Table 100 summarizes all findings not directly related to the OT analyses Because of the nature of the analyses we used mostly odds ratios and testsit was necessary to assign individuals to either exposed or unexposed group regardless of which of the chemical groups was being investigated The central limiting factor however in all investigations conducted at the OHSG and indeed in much of occupational research concerns the validity of these exposure classifications which often involve arbitrary decisions While these arbitrary decisions are certainly based on reasonable approximations to the truth it becomes necessary to make rather extensive generalizations and assumptions There is also much room for subjectivity and error and both of these factors may lead to misclassification and loss of validity Exposure classification for all analytic studies conducted to date at the OHSG is based the Occupational Title OT system This system condenses the thousands of jobs within the rubber industry into approximately 100 OTs which share common production processes It should be noted that this system does not claim to be based on specific exposures although the assumption is often made that common production processes can be equated with common exposures This equation is at JACKSON 025697 257 TABLE 100 SUMMARY OF MAJOR FINDINGS NOT DIRECTLY RELATED TO EXPOSURES Marital Status Slightly higher proportioonf separated widowed or divorced in Company A no difference in Company B. Education No significant differences in either company Birthplace Slightly greater proportion of cases were foreiingean ch company with a greater discrepancy in Company B. Calendar Year of Hire Earlier year of hire for Company A no difference for Company B. Difference in Company A confined to comparison of cases with matched controls Age at Hire No difference for Company B slightly younger age at hire for cases in Company A with difference confined to comparison of cases with matched controls Total Duration in Active OTS No difference for CompanBy slightly longer duration for Company A cases with difference confined to ARSD comparison JACKSON 025698 258 best moot We note however that even two individuals performing the same job can differ considerably in their exposure to toxic substances by virtue of the precise manner in which they perform their jobs much as two smokers who consume the same number of cigarettes may differ considerably in the amount of carcinogen they inhalbey virtue of their style of smoking The difficulty in assessing exposures and exposure levels therefore begins at the level of the individual and is compounded at each step of the decision process which results in ever broader classifications At the OT level we may be dealing with category which includes perhaps fifty different jobs in more than one area of the plant with obvious inherent difficulties regarding assumptions that all individuals in this OT are subject to the same exposures These difficulties are further compounded by the next categorization which involves condensing the 100 OTs into a more manageable number of perhaps twenty which are more easily handled in analyses What we have therefore is a system which requires at every step ever broader classification schemes and ever more general assumptions which may be drifting further and further from truth all of which may lead to serious loss of validity To further confuse the issue we assume in our system that a surrogate for dose is estimated by duration of employment within OT already having assumed that all individuals have the same exposure dose for one days work and that this exposure has not temporally changed so that whether the exposure occurred in 1930 or 1970 they are qualitatively and quantitatively similar JACKSON 025699 259 Despite these limitations it must immediately be added that the : system in use at the OHSG is probably as good as any other available for hypothesis generation in occupational studies After all any large industry demands the performance of very many diverse jobs and the investigator is faced with the problem of condensing this large number into a small based on the number of meaningful subgroups the classification being particular investigation being conducted While the creation of these groups is difficult enough even when a specific hypothesis is being tested it becomes overwhelmingly difficult when the investigator is engaged in a generating endeavor where he essentially admits that he is uncertain about the nature of the agent + which he believes is responsible for an excess of some disease In this investigation both approaches hypothesis testing and generating were used although the study should be considered mainly of the hypothesis- generating type For the hypothesis generating used twenty more or less standard OTGs used portion of the study we in other investigations with the hope of uncovering some high risk area which could then b more fully explored in terms of specific exposures within these areas Because of the possible existence within the industry of four specific agents or classes of chemicals nitrosamines polycyclic hydrocarbons talc containing asbestos and carbon black implicated as possible etiologic factors specific OTGs for in the development of these chemicals For gastric malignancies we created each chemical we created four QTGS with potential for high moderate low and no exposure based on an ordinal exposure scale ranging from 0 to 3. Each of three industrial JACKSON 025700 260 scientists at OHSG Messrs Arp Gamble and Williams independently evaluated each OT and determined where each of these four classes of | chemicals would have been used or generated within the industry if they were present at all and this was done for the broad time frame 1915-1970 based on which an exposure score was assigned These scores were then averaged following which the exposure categories were created and analyzed We have therefore a system of exposure classification which while imperfect nevertheless has a reasonable observer reliability and can be considered an approximation to definitive exposure indices Now that we have considered the validity of the exposure classifica- tion or external validity we should turn our attention to the question of internal validity or the validity of the specific comparison of cases with controls and of the conclusion based on these comparisons The essence of internal validity involves the unbiased nature of the estimate of effect whicihs said to be biased from lack of internal validity if the estimate is inaccurate because one or the other of the comparison groups has certain attributes which can affect the outcome and which are differentially distributed between the groups That is the outcome is being affected in a differential way by this attribute so that the estimate of effect which is based on an internal comparison e.g. rate , among exposed vs. rate among the unexposed is no longer accurate or valid this being an issue of internal validity The attribute referred to is generally called a confounding factor because its existence confounds or obscures the relationship of interest destroys internal validit anyd leads to a biased estimate of effect JACKSON 025701 261 A confounder can be conceptualized as a predictor or alternate cause of the outcome under study and it must be differentially distributed among the groups being compared for the estimate of effect to be biased This can be presented diagramatically as S \ Exposure Disease . In other words a factor is a confounder if it is distributed differently among the groups being compared exposed and unexposed while it is predictive of the disease in each group separately or independent of exposure The former condition is represented by a squiggle while the latter is represented by an arrow For instance if one conducts a ten year lung cancer cohort study where members of the exposed group are all young while the unexposed group is older we can expect to find ahigher lung cancer rate among the unexposed than would otherwise have occurred simply because they are older than the exposed group and age is predictive of lung cancer among both smokers and smokers Unless this difference is controlled for the estimate of the effecotf smoking on the incidence of lung cancer will be biased due to confounding by age The most widely used methods for the control of confounding include matching in the design stratification on the confounder in the analysis or use of multivariate analytic techniques such as multiple regression or analysis of covariance Investigators often assume that certain factors most commonly age race and sex are universal confounders meaning that they will confound any investigation undertaken To prevent this potential loss JACKSON 025702 262 of internal validity therefore subjects are matched on these characteristics during the subject selection phase of the study Methodologists often stress however despite this universal assumption that a factor isa confounder only if it is a confounder in the data for the specific study being conducted If one accepts this admonition in | practical way one would never match during the design of any study We on the other hand felt thaitt was reasonable to assume that differences in the age race and sex of subjects would lead toa biased estimate of effect in our study since studies of stomach cancer | it did so in virtually all | reviewed In additiotno matching on these three attributewse matched one of the control groups ARSD on total duration of employment as well because we felt we had to provide equal opportunity for all subjects to have worked in all OTs If we had not provided this equal opportunity it could have been argued on finding an elevated odds ratio that the | explanation lies not with some related exposure but that the elevated ratio could be function of the greater length of time spent by . the group under study within the industry To circumvent this additional problem in data interpretation therefore we matched one group on this factor using the difference between the seniority date and the last date worked as an estimate of total duration of employment When a determinatiwoans made however , employment within active OTs for each of the of the total duration of two control groups there was no statistically significant difference between the two control groups in this modified duration In essence therefore we had four controls JACKSON 025703 263 for each case matched either directly or indirectly on age race sex and duration of employment in active OTs The absence of any difference between the two control groups is probably explained on the basis of the age match most workers within the rubber industry begin their employment at a relatively young age approximately 30 and spend most of their productive lives within the industry Having matched on age year of birth and age at death therefore we have apparently indirectly matched on duration of employment as well This finding may however also be explained by the long latency of the disease and the common practice of ~ working until retirement age with the combination of these two factors responsible for the similar durations of employment and this would have occurred even without matching Because the two control groups had similar total active durations and these durations did not significantly differ from the durations of the cases the original plan calling for the termination of the work history of controls when they had attained a duration equal to that of the case was ignored For most analyses therefore data from all controls were pooled and no artificial termination of work experience was executed While those similarities in duration simplified the analyses however it should be borne in mind that we lost of a generalized effect which would the opportunity to have been evidence test the hypothesis by a difference in duration between cases and comparands Having matched indirectly on duration the opportunity to test this hypothesis disappeared While on this subject of indirect matching it should also be noted that there were no consistent statistically significant differences between cases and JACKSON 025704 264 controls for year of hire and age at hire and this also may have been caused by indirect matching At this point a word about the analysis is in order Because we avoided many confounders by matching during subject selection we were forced to utilize specific analytic procedures such as Haenszel Odds Ratios and Paired Tests which maintain the integrity of the matched sets or stata Referring back to the diagram of a confounder we see that since the confounder is itself a predictor or alternate cause of the disease or outcome the exposed and unexposed groups can be expected to have because they were matched on numerous true confounders outcomes which are more similar to one another than would otherwise have occurred had they not been matched on these factors That is matching on true confounders decreases the difference in outcome between the two groups by reducing the variance in the measure of effect and hence increasing the precision of this estimate As a consequence matched analyses can be conceptualized although not in strictly mathematical sense as procedures which take into account this smaller difference thereby permitting the conclusion of a statistically significant difference between the groups being compared even though this differenmcaey be smaller than would otherwise be necessary for concluding that a significant . . difference exists While use of an appropriate analytic procedure maintains the integrity . of the matched sets however there are fewer of those procedures available and they are often computationally more difficult to use For this reason alone investigators should carefully consider all aspects of this JACKSON 025705 265 study before making a decision regarding matching in the design Alternative methods for controolf confounding such as stratification in the analysis which was used for numerous confounders in this study or subject selection restriction which was not used should be extensively weighed before any conclusions are made since an ideal design is not much use without the appropriate techniques for analysis especially when the questions being asked are rather complicated and require sophisticated analytic procedures While it is obvious from reading this report that most of the analyses presented maintained the matching we did occasionally avail ourselves of unmatched analyses For instance in examining the odds ratios separately for foreign and native born cases it would be productive to maintain the matching since matched controls may or may not have the same birthplace as the case which means that we might be comparing odds ratios for foreign born cases and native born controls or vice versa which would defeat the purpose of the analysis in which instance dissolution of the matching would be the more appropriate analytic technique At this point before discussing the results of this investigation it would be in order to present some possible limitations aside from exposures study the difficulty of determining true _ of this First consideration must be given to the fact that case identification for this study relied upon mortality from and not the incidence of stomach cancer While we have great confidence in our follow of the large cohorts studied at OHSG approximately 95 and despite the fact that JACKSON 025706 266 | the median survival time for stomach cancer is less than half a year the possibility nevertheless exists that we have systematically missed the younger and less severe cases with the greatest chance for survival especially but not exclusively since the cohort from which the cases were derived required that all members be forty years of age or greater at inception of the study These cases if they exist may indeed be qualitatively different and may have had different exposures from those which were included in this study In addition to this possible limitation concerning cases it should also be noted that there was no check of the hospital records of controls so the exclusion of individuals as eligible controls because of malignancies was based entirely on the conditions listed on the death certificates Indeed one may even question the validity of excluding all individuals with malignancies from the comparand pool since there is certainly no evidence that all malignancies are associated with the same exposures as those believed to be responsible for the excess mortality from gastric cancer And one must also consider that by excluding malignancies from eligibility we have forced most of the deceased controls to have died from either heart disease or stroke the two conditions which along with cancer are responsible for the great majority of deaths in this country And if those controls who died of cardio- vascular diseases happen to have exposures similar to those with stomach cancer for whatever reason we are faced with estimates of effect exposure biased towards the null and hence loss of validity Restrictions regarding control selection therefore are and will JACKSON 025707 267 probably always remain debatable regardless of which criteria are employed Three final brief points should be made at this time regarding this project study First while there was a considerable effort expended to determine the economic status SES of subjects in this study from evaluations of each job by the industry on which their pay scale . was at least partly based there were problems with this analysis which precluded its inclusion in the present study While we do not expect much variation in SES among collar workers and analysis of educational attainments did not reveal any differences between cases and controls this factor is important in the study of stomach cancer Second there is evidence that the relative risk of developing stomach cancer smokers compared with smokers is approximately 1.3 and the same may be true for alcohol consumption Since we were unable to determine smoking and drinking histories the possibility exists that there was a higher proportion of smokers or drinkers among cases which could explain part of the excess Third the nativity factor is of major importance in the study of stomach cancer and even second and perhaps third this is just now being extensively studied among migrants from Japan generation Americans born of parents or grandparents from high risk countries continue to experience excess mortality from . stomach cancer Since we were only able to determine the nativity of the subjects included in this investigation and not the nativity of their forebearers there may have been differences which contributed to the excess disease JACKSON 025708 268 Occupational Associations The most consistent finding in this study is the very general observation that there is not much difference in either occupational or occupational factors between cases and comparands in Company B while there are some differences in the comparisons of Company A cases and controls We found among Company A subjects that cases had worked in active OTs slightly longer than controls 375 vs. 338 months , and this difference was significant in the comparison with one ARSD of the two control groups We also found that cases started at a slightly 1.5 years younger age and slightly earlier 1.5 years than their matched controls this latter difference being statistically significant for the comparison with one of the control groups although most of these differences were traced to the group of cases entering during the first decade of plant operation The differences between Company B cases and controls were generally less pronounced and not statistically significant as well although slight differences often were found In addition to these factors a higher proportion of Company A cases were foreign although a greater proportion of controls were foreign born as well . | The exposure analysis of occupational classification systems - exposures one using using was based on two different standard OTGs based on similar aspects of production and not necessarily on common exposures, and the second based on the potential for specific chemical exposures a system developed exclusively for this study and based on a review of the literature including both human and animal studies from which were JACKSON 025709 269 derived a list of chemicals - aromatic hydrocarbons nitrosamines asbestos and carbon black the latter possibly being a carcinogen in its own right or a particulate to which the polycyclics may be absorbed whose association with stomach cancer appeared regularly in the literature and which were used as raw materials or generated by industrial . processes within the industry The creation of the specific high risk OTGs as described earlier in this chaper was based on three independent assessments of the levels of these chemicals within each of the OTs and those assessments were made by the three OHSG environmentalists most knowledgeable about these matters While they are based on extensive experience they must in the end be considered at least partially subjective to varying degreeass will soon be explained and they do not escape the limitations of the OT system The reference to varying subjectivity must be further discussed here One of the groups the nitrosamines are not used as raw materials per se and an estimate of their probable levels is therefore based on judgement concerning where they might be generated although in principle one can sample for a precise determination of levels Aromatic hydrocarbons , as well are not used in raw form although they may be contaminants of some commonly used solvents Considerable knowledge is presently available concerning the use of these solvents within different OTs The determina- tion of carbon black levels is rather direct because we know exactly what it does and where it is used in addition to being visible Assessment of asbestos levels on the other hand are indirect because they are based on the presence and use of sticking detackifying JACKSON 025710 270 agents which are most often talc or soapstone although other agents such as mica or even cornstarch are also used depending on costs at the time of purchase The most used agent however is believed to be talc and this substance is believed to contain asbestiform material to varying degrees Despite the assumption on which asbestos determinations are based however talc like carbon black is visible and its use within the industry is reasonably well documented An examination of the results from the standard OTG classification system did not reveal any pattern of consistently elevated odds ratios with most elevations being generally scattered throughotuhte earlier end of the industrial process and being found in the following OTGs Calendering Tuber Tire and Bead Fabrication Tube Fabrication Cure Preparation and Curing While there was no striking consistency to these findings it is interesting to note that these OTGs generally have high dust levels consisting mostly of talc or other detackifiers and also use solvents extensively the latter being possibly contaminated with polycyclic hydrocarbons The most provocative finding occurred when using the chemical- specific OTG classification which indicated significantly elevated odds ratios for potential exposure to high or moderate talc levels with the ratios being significantly elevated for exposure on an never basis as well as for exposure of two five and ten years JACKSON 025711 271 duration each compared with no exposure at all While these statistically significantly elevated ratios are not independent of one another since subjects with ten years of exposure are included in the other three groups the consistency was nevertheless striking This finding pertains only to Company A however and the similar analysis conducted within Company B indicated interestingly that the odds ratios were significantly low a finding which may be attributed to random fluctuations and not to some underlying protective elements within these OTs Equally and perhaps more likely may be selective migration from dusty areas a reasonable explanation only if company policies and | | practices differed in regard to selective job changes In addition to these findings of statistically significantly elevated odds ratios within Company A we have additional evidence for the association of stomach cancer with OTs high in probable talc or detackifier exposures from the Test utilizing durations of employment within these OTs which indicated significantly more time forty versus twenty months ) spent in these OTs by the cases and the additional finding that the exposures among the cases occurred earlier than it did among the controls when dust levels may have been higher Since there was a question concerning the possible causal effects of the slightly younger age and earlier calendar year of entry into the industry for cases in comparison with matched controls separate Tests and Odds ratios for asbestos exposure were calculated for this group as well as for the matched controls The Tests results described in the previous chapter indicated virtually no difference in duratioonf exposure for the two groups and the ORs were similar JACKSON 025712 272 The combination of those findings demonstrates the existence of an association between stomach cancer and dusty areas in which talc may have been used and which could have been ingested via capture by mucus in the nasopharynx or by expectoration and subsequent swallowing a mechanism postulated by numerous investigators for the ingestion of particulates Because of the earlier finding of a disproportionate number of early entry cases separate analyses by decade of entry and by latency period were performed the latter based on the difference between year of death and year of hire on the assumption that level exposures began during the first year within the industry These analyses were unmatched and indicated that the odds ratios for potential asbestos exposure were elevated for the earlier decades and longer latency periods with the OR for a latency of 40 or more years significantly elevated OR = 2.24 while the ratio for entry prior to 1930 OR = 2.61 was also significantly elevated It appears therefore that mucohf the excess can be statistically attributed to those individuals who began working in the industry quite early at which time the dust and hence potential asbestos exposure may have been higher than the levels during later years This is further confirmed by the fact that the group with the 40 year potential latency first entered a high talc OT on the average in 1930 while controls with a similar potential latency had a mean first entry year of 1937. The medians for these two groups were 1926 and 1936 respectively From an examination of Table 97 we observe a clearly decreasing gradient of mean exposure among exposed cases but this same gradient does not exist among controls In fact only among those with JACKSON 025713 273 the longest potential latency do we observe longer exposures among cases From these decade of entry and latency analyses therefore it appears that the excess of deaths from stomach cancer is attributable to a combination of earlier and longer exposures The attributable risk per cent among the exposed OR multiplied by the proportion of cases exposed for this exposure using an odds ratio of 2.5 and a case exposure proportion of fifty percent 27/52 is calculattoebde thirty percent which indicates overall that the occurrence of thirty percent of the fifty cases or approximately sixteen cases can be attributed to detackifier exposure assuming no bias or random measurement errors Similar calculations are not presented for Company B naturally since no association was demonstrated In view of the fact that the point estimates of the SMR on which these studies are based were hundred eighty and hundred twenty for Companies A and B respectively and considering that the published Company A SMR 135 was presented as being statistically significantly elevated while the published Company B SMR 136 was not it appears that the absence of a statistically significant excess in Company B is confirmed by the absence of any consistent findings for that company in this study while the signifexicecssaanmotng Company A subjects is confirmed by our analysis and may be largely explained by a history of working in areas with probable exposure to sticking agents mostly talc Because of the presumed contamination of talc with asbestos especially in earlier calendar periods there is a fairly . strong possibility that this involved asbestos exposure as well In Company A we observed fifty cases and expected twenty and JACKSON 025714 274 of the twenty excess deaths sixteen appear to be attributable to employment within OTs thought to contain high levels of asbestos These findings therefore appear to explain most of the observed excess Suggestions for Further Research First consideration for further research must be given to the creation of a classification system based directly on exposures Such a system is presently being developed here at the OHSG and is being based ona grid system in which workers in close physical proximity to each other are grouped together While this system is very promising it will probably not be appropriate for retrospective use which means that the present OT system is the only and therefore the best available Since we are now entering the surveillance phase of our activities however every effort should be made to refine this new system as quickly as possible The data on which our conclusions are based in this present study would doubtless have been better if we had knowledge concerning the actual dose of exposure and not just the amount of time spent in the various OTs If these levels can be reconstructed perhaps from purchasing records , analysis with this additional factor would be appropriate It would be especially interesting to compare samples of Company A and B talc or detackifiers collected when most exposures in this study 1915-1950 1915-1950 occurred or in lieu of this to trace the suppliers and from this to reconstruct the asbestos levels within these products JACKSON 025715 275 In addition to further analysis of the asbestos levels it would be appropriate to further examine these data for a determination of the historical periods during which both cases and controls were exposed to solvents since there is some suggestion in our data concerning an association between stomach cancer and solvents and since the purity of the solvents and hence possible contamination by polycyclic hydrocarbons apparently changed significantly around 1940. Since expectoration and subsequent ingestion or via the mucus system is the presumed mechanism it would be very interesting to examine those OTs associated with both chronic pulmonary diseases and high asbestos levels since the former may cause a greater frequency of expectoration and hence of ingestion It would also be desirable to perform a more detailed analysis of the specific exposures which occurred during the estimated latency period twenty years for stomach cancer although the variability in the estimate of this latency period is great and the period itself is long Finally in view of the many factors both occupational and nonoccupational which may be associated with the incidence of stomach cancer such as nativity economic status and age and calendar year at hire among others it would probably be best to control for all of these factors simultaneously by the use of an appropriate technique such as perhaps the confounder summarization score JACKSON 025716 REFERENCES JACKSON 025717 277 REFERENCES McMichael AJ Spirtas R Kupper LL An epidemiologic mortality within a cohort of rubber workers 1964-72 Med 16 458-64 1974 study of J Occ End Results in Cancer Report No. 4. S. Dept of Health Education and Welfare National Cancer Institute 1972 | Doll R The geographical distribution of cancer 23 1-8 1969 Brit J Cancer Dunham LJ Bailar JC III World maps of cancer mortality rates and frequency ratios J Nat Cancer Inst 41 155-203 1968 U. S. Cancer and Welfare 1975 Mortality by County U. S. Dept Public Health Service National of Health Education Institute of Health Sterup K Mosbech J Sex ratios of gastric cancer related to site of the tumor Scand J Gastroen Suppl 9 87-89 1971 MacDonald WC Clinical of the gastric cardia and pathological features Cancer 29 724-32 1972 of adenocarcinoma Flamant R Lasarre 0 Lazar P Lequerinais J Denoix P Schwartz D Differences in sex ratio according to cancer site and possible relationship with the use of alcohol and tobacco Review of 65,000 cases J Nat Cancer Inst 32 1309-16 1964 10. Griffith GW The sex ratio in gastric considerations relative to aetiology 1968 cancer Brit J and hypothetical Cancer 22 163-72 Haenszel W Variations cancer with particular Cancer Inst 21 213-62 in incidence reference to 1958 of mortality from stomach the United States Nat 11. Levin ML Haenszel W Carrol BE Gerhardt PR Hardy VH Ingraham SC III Cancer incidence in urban and rural areas of New York State J Nat Cancer Inst 24 1243-57 1960 12. 13. 14. 15. Hoover R Mason TJ McKay FW resource for etiologic clues Fravmeni JF Science 189 Cancer by county 1005-07 1975 new Gregor 0 mortality 1974 Toman R Prusova F Relation of gastrointestinal to mortality in general Scand J Gastroent Suppl cancer 9 79-85 Newell GR Waggoner DE Cancer mortality and environmental temperatures in the United States Lancet 1 766-68 1970 Davies RI Wynne Griffith G Cancer and soils in the county of Anglessey Brit J Cancer 8 56-66 1954 JACKSON 025718 278 16. Cohart E Socioeconomic distribution of stomach cancer in New Haven Cancer 7 455-61 1954 17. Graham S Levin M Lilienfeld of cancer of various sites in 180-91 1960 AM The Buffalo socioeconomic distribution N. Y. 1948-52 Cancer 13 ; 18 19 20. Seidman H Cancer death rates by site socioeconomic groups in New York City 1970 and sex for religious and Environ Res 3 234-50 Newill VA Distribution of cancer mortality among ethnic subgroups of the white population of New York City 1953-1958 J Nat Cancer Inst 26 405-17 1961 : Buell P Dunn JE Jr Breslow L The occupational class risks of cancer mortality in men J Chron Dis 12 600-21 1961 21 Kraus AS Levin ML Gerhardt PR A study of occupational with gastric cancer Am J Pub Hith 47 961-70 1957 associations 22. Sigurjonnson J Occupational cancer in relation to dietary 1967 variations in differences mortality from gastric Br J Cancer 21 651-56 23. Lombard HL Doering CR Cancer studies in Massachusetts III Cancer mortality in nativity groups J Prev Med 3 343-61 1929 24. Choi NW Ethnic distribution of cancer of the gastrointestinal tract in Manitoba Amer J Pub Hith 58 2067-81 1968 25. Graham S Levin ML Lilienfeld AM Sheehe P Ethnic derivations as related to cancer at several sites Cancer 16 13-27 1963 26 Haenszel W Cancer mortality among foreign born in the United States J Nat Cancer Inst 26 37-132 1961 27. 28. Smith RL Recorded and expected mortality the United States and Hawaii with special Nat Cancer Inst 17 459-73 1956 among the reference Japanese of to cancer Buell P Dunn JE Jr Nisei of California Cancer mortality among Japanese Issei and Cancer 18 656-64 1965 29. Staszewski J Haenszel W Cancer mortality among the Polish born in the United States J Nat Cancer Inst 35 291-97 1965 30. Staszewski J McCall MG Stenhouse NS Cancer mortality in 1962-66 among Polish migrants in Australia Brit J Cancer 25 599-610 1971 31. Monk M Warshauser ME Stomach and colon Puerto Ricans in New York City and Puerto 28 349-53 1975 cancer Rico mortality among J. Chron Dis JACKSON 025719 279 .32 Tulchinsky D Modan B Epidemiological aspects of cancer of the stomach in Israel Cancer 20 1311-17 1967 33 Terris M Hall CE Decline in mortality from native and foreign residents in New Cancer Inst 31 155-62 1963 gastric cancer in York City J Nat 34 Mancuso TF Coulter EJ MacDonald EJ Migration and cancer mortality experience - a nonwhite Ohio residents study of native and southern born Trace Subs in Env Hith : 49-56 1973 _ 35. Bjelke E Epidemiologic studies of cancer of and rectum with special emphasis on the role Gastroen Vol 19 Suppl 31 1974 the stomach colon of diet Scand J . 36. Shamberger RJ Tytko S Willis CE Antioxidants in cereals and in food preservatives and declining gastric cancer mortality Clev Clin Quart 39 199-24 1972 37. 38. Correa P Haenszel W Cuello C Tannenbaum S Archer M for gastric cance epidemiology Lancet 58-59 1975 A model Stocks P Cancer of the stomach in the large towns of England and Wales 1921-39 Br J Cancer 4 147-57 1950 39. Aird I Bentall HH Roberts JAF A relationship between cancer of the stomach and ABO blood groups Br Med J : 799-801 1953 40 Buckwalter JA Wohlwend CB Colter DC Tidrick RT The association of the ABO blood groups to gastric Surg Gynecol Obstet 104 176-79 1957 Knowler LA carcinoma 41. Billington BP Gastric site and epidemiology cancer - relation between Lancet 2 859-62 1956 ABO blood , groups 42. White C Eisenberg H ABO blood groups and cancers of the stomach Yale J Biol Med 32 58-61 1959 43 Crumb CK Willets PF Stephenson HE Gastric cancer - a descrescent malignancy Surgery 68 277-82 1970 44 Flamant R Lasserre 0 Lazar P Lequerinais J Denoix P Schwartz D Differences in sex ratio according to cancer site and possible relationship with use of tobacco and alcohol review of 65,000 cases J Nat Cancer Inst 32 1309-16 1964 _ 45 Schwartz D Flamant D Lellouch J Denoix P Results of a survey on the role of tobacco particularly inhalation on cancer sites J Nat Cancer Inst 26 1085-1108 1961 French different ; 46 Hammond EC Smoking in relation to mortality and morbidity Findings in first thirty months of follow in a prospective study started in 1959. J Nat Cancer Inst 32 1161-88 1964 JACKSON 025720 280 47. Wynder EL Conference on the etiology of cancer of the gastro- intestinal tract Cancer 19 1561-66 1966 48 Zacho A CederquisCt Nielsen J Larsen and quantity of tobacco used by patients Acta Med Scand 193 45-48 1973 V Duration with gastric of smoking cancer 49 50 Zacho A Nielsen J Larsen V Smoking habits of patients with gastric cancer Acta Chir Scand 137 455-60 455-60 1971 Stocks P Cancer mortality in relation to national of cigarettes solid fuel and coffee Brit J 1970 consumption Cancer 24 215-25 51 Graham S Schotz W Martino P Alimentary factors in the epidemiology of gastric cancer Cancer 30 927-38 1972 52 53. Davies RI Wynne Griffith G Cancer and soils in the county of Anglesey Brit J Cancer 8 56-66 1954 Tromp SW Diehl JC A statistical study between cancer of the stomach and soil 1955 of the Brit J possible relationship Cancer 9 349-57 54 Stocks P Davies RI Zinc and copper content of soils associated with the incidence of cancer of the stomach and other organs Brit J Cancer 18 14-24 1964 55. Armijo R Coulson AH Epidemiology of stomach cancer in Chile the role of nitrogen fertilizers Int J Epid 4 301-09 1975 56. Boyd JT Doll R Gastrointestinal cancer and the use of liquid parrafin Brit J Cancer 8 231-37 1954 57 Graham S. Lilienfeld AM Tidings JE Dietary and purgation factors in the epidemiology of gastric cancer Cancer 20 2224-34 1967 58 Schonland M Bradshaw E Indians with reference to 23 670-82 670-82 1969 Upper alimentary tract cancer in the chewing habit Brit J Natal Cancer 59 Elliot CV Wald SM Benz RI A roentgenologic study of ulcerating lesions of the stomach Amer J Roentgen 77 612-22 1957 60 Cohn I Jr Sartin J Sudduth P J Gastroent 32 121-35 1959 Giant ulcers of the stomach ; Amer 61 Hirohata T Kuratsune M The geographical comparison of mortality from cancer of the stomach and ulcer of the stomach in Japan Br J Cancer 23 465-79 465-79 1969 62 Bera JW Histologic aspects of the relation between gastric adenomatous polyps and gastric cancer Cancer 11 1149-55 1958 JACKSON 025721 281 63. Ming SC Goldman H Gastric polyps a histologic classification and its relation to carcinoma Cancer 18 721-26 1965 64. Tomasulo J Gastric polyps histologic types and their relationship to gastric carcinoma Cancer 27 1346-55 1971 65 Monaco AP Roth SI Castleman B Welch CE Adenomatous polyps of the stomach a clinical and pathological study of 153 cases Cancer 15 456-67 1962 66 Marshak HH Feldman F 1965 Gastric polyps . Am I Dig Dis 10 909-35 a 67. Comfort MW Kelsey the development of 7 367-73 .1947 MP Berkson J Gastric acidity before and after carcinoma of the stomach J Nat Cancer Inst 68. Stempien SJ Rodiles DH Cooper EJ Dagrad AE Clinical correlation with quantitative gastric acid measurements Am J Gastroent 61 261-69 1974 69 - 70. Pastor J Kato H Belsky J Serum pepsin and tubeless gastric analysis as predictors of stomach cancer NEM 286 279-84 Vatn MH Gastric intrinsic factor secretion 10 337-38 1975 Scand J Gastroent 71. 72. 73 Sivralla M Seppala K Atrophic gastritis as a possible precursor of gastric carcinoma and pernicious anemia results of follow follow examinations Acta Med Scand 166 455-74 1960 Kaplan HS Rigler LG stomach Am I Med Sci Pernicious anemia and 209 339-48 1945 carcinoma of the Morson BC Intestinal metaplasia of the gastric mucosa Cancer 9 365-76 1955 Br J 74 75. 76 77 78 Morson in the BC Carcinoma arising from areas of intestinal gastric mucosa Br J Cancer 9 377-85 1955 metaplasia Ming S Tumorosf the Esophagus and Stomach of Pathology Washington DC 1973 Armed Forces Institute Stout AP Tumors of the Stomach Pathology WashingtoDnC 1953 Armed Forces Institute of carcinoma Lauren P The two histologic main types of gastric diffuse and so called intestinal type carcinoma Acta Path aan Microbiol Scand 64 31-49 1965 Munoz N Collelly of gastric cancer 1971 :. Time trends of intestinal and diffuse types in the United States Int J Cancer 8 158-64 | JACKSON 025722 282 79. Munoz N Correa P Cuello C in high- and risk areas Histologic types of gastric carcinoma Int J Cancer 3 809-18 1968 80 Crane PS Rhec of the stomach 747-51 1970 SV Scel DJ seen in Korea Experience with 1079 cases from 1962 to 1968. Amer J , of cancer Surg 120 81. Waterhouse JAH Cancer Handbook of Epidemiology and Prognosis Churchill Livingstone Edinburgh and London 1974 82. Hopps HC Principles of Pathology York 1964 Appleton New : 83 84. Robbins SL Angell M Philadelphia 1971 Basic Pathology W. B. Saunders Co 7 Miller JA Carcinogenesis by chemicals Research 30 559-76 1970 an overview Cancer 85 Kellerman G Shaw CR Kellermann M Aryl hydrocarbon hydroxylase inducibility and bronchogenic carcinoma NEJM 289 934-36 1973 86 Winkelstein W Jr Kantor S Stomach cancer - positive association with suspended particulate air pollution Arch Environ Hlth 18 544-47 1969 87 Matolo cancer NM Gorishek WM of the stomach Moslander V Dixon JA Coal Rocky Mount Med J 69 44-49 mining 1972 and 88. Creagan ET Hoover RN Fraumeni cancer in coal mining regions JF Jr Mortality Arch Environ Hith from stomach 28 28-30 1974 89. Stocks P On the relations between atmospheric pollution in urban and rural localities and mortality from cancer bronchitis and pneumonia with particular reference to 4 benzopyrene berylium molybdenum vanadium and arsenic Br J Cancer 14 397-418 1960 90. Magee PN Barnes JM tumors in the rat by 114-22 1956 The production of malignant feeding dimethylnitrosamine primary hepatic Br J Cancer 14 91. Hill MJ Hawksworth G Tattersall G Bacteria nitrosamines and cancer of the stomach Br J Cancer 28 562-67 1973 92. 93. Magee PN Possibilities of hazard from nitrosamines in industry Ann Occ Hyg 15 19-22 1972 Low H Nitroso compounds Hlth 29 256-60 1974 - safety and public health oe Arch Environ | 94. Merliss RR treated rice and Japanese stomach cancer 173 1141-42 1971 Science JACKSON 025723 283 95. Kleinfeld M Messite J Zaki MH Mortality experience among talc workers a follow study J Occ Med 16 345-49 1974 96. Henderson WJ Evans DMD Davies JD Griffiths K Analysis of particles in stomach tumors from Japanese males Env Res 9 240-49 1975 97. Lloyd JW Ciocco A Long mortality study of steelworkers I. Methodology JOM 11 299-310 1969 98 99. Robinson H Long mortality study of steelworkers II Mortality by level of income in whites and whites JOM 11 411-16 1969 Redmond CK Smith EM Lloyd JW Rush HW Long mortality study of steelworkers III Follow JOM 11 513-21 1969 100. Lloyd study 1970 JW Lundin Jr. FE Redmond CK Geiser PB Long of steelworkers IV Mortality by work areas JOM mortality 151-57 101. Lloyd JW Long mortality study of steelworkers cancer in coke plant workers JOM 13 53-68 1971 V. Respiratory 102. Redmond CK Ciocco A Lloyd JW Rush HW term mortality study of steelworkers VI Mortality from malignant neoplasms among coke oven workers JOM 14 621-29 1972 . 103 104 105. Lerer TJ Redmond CK mortality study among crane operators JOM Breslin PP Slavin L Rush HW steelworkers VII Mortality 16 608-14 1974 term patterns among Lerer TJ Gustin J Kamon E term mortality study steelworkers VIII Mortality patterns of open hearth a preliminary report JOM 17 40-43 1975 among steelworkers . Mazumdar S Lerer T Redmond CK Long mortality study among steelworkers IX Mortality patterns among sheet and tin mill workers JOM 17 751-55 1975 106. 107. 108. Lynch KM Smith WA Pulmonary asbestosis III - carcinoma of lung in asbestosis Amer J Cancer 24 56-64 1935 Wagner JC Sleggs CA Marchand P Diffuse and asbestos exposure in the Western Ind Med 17 260-71 1960 pleural mesothelioma Cape Province Br J Doll R Mortality from lung cancer in asbestos workers Ind Med 12 81-86 1955 Br J 109. Selikoff IJ Churg J Hammond EC The occurrence of asbestos among insulation workers in the United States Ann N. Y. Acad of Sc Bt 132 139-55 1965 JACKSON 025724 284 110 Selikoff IJ Churg J Hammond EC Asbestos exposure and neoplasia JAMA 188 22-26 1964 111 112 113. 114. 115. Enterline PE Mortality among asbestos products workers in the United States Ann N. Y. Acad Sci 132 156-65 1965 Braun DC Truan TD An epidemiologic study of lung cancer in asbestos miners Arch Ind H1th 17 634-52 1958 Selikoff IJ Hammond EC Seidman H Carcinogenicity of amosite asbestos Arch Env Hlth 25 183- , 1972 Meurman LO Kilviluoto P Hakama M employees of anthophyllite asbestos Med 31 105-12 Mortality and morbidity of mines in Finland Br Br J Ind McDonald JC McDonald AD Gibbs GW asbestos mines and mills of Quebec Mortality in the chrysolite Arch Env Hlth 22 677-86 1971 116. Stell PM McGill T Asbestos and laryngeal carcinoma Lancet 2 117. Shettigara PT Morgan RW Asbestos smoking and Taryngeal carcinoma Arch Env Hlth 30 517-519 1975 118. Lorenz E Radioactivity lung cancer in miners of Inst 5 1-15 1944 and lung cancer a critical Schneeberg and Joachimsthal review of J Nat Cancer 119. Archer VE Magnuson HJ Holaday DA Lawrence PA Hazards to health in uranium mining and milling JOM 4 55-60 1962 120 Wagoner JK Miller RW cancer mortality among 269 284-89 1963 Lundin FE a group of Fraumen JF Haij ME Unusual underground metal miners NEJM 121. Wagoner JK Archer VE Lundin FE Radiation as the cause of lung cancer among uranium miners JEJM 273 181-88 1965 122 Wagoner JK Archer VE Carroll BE Holaday DA Lawrence PA Cancer mortality patterns among U. S. uranium miners and millers 1950 . through 19 . J. Nat Cancer Inst 32 787-801 1964 123. Saccomanno G Archer VE Saunders RP Lung cancer of uranium miners - on the Colorado Plateau Hlth Phys 10 1195-1201 1964 124. Archer VE Saccomanno G Jones JH types of bronchogenic carcinoma as Cancer 34 2056-60 1974 Frequency related to of different histologic radiation exposure. 125 Archer VE Wagoner JK Lundin FE Lung cancer among uranium miners in the United States Hlth Phys 25 351-71 1973 JACKSON 025725 285 126. 127. Selikoff IJ Hammond EC Churg J Asbestos exposure smoking and neoplasia JAMA 204 104-112 1968 Lundin FE Wagoner JK Archer UE Radon daughter exposure and respiratory cancer - quantitative and temporal aspects U. S. Dept of HEW PHS NIOSH and NIEHS Joint Mangr No. 1 1976 128 Donaldson AW Epidemiology of lung cancer among uranium miners Hlth Phys 563-69 1969 129 World Health Statistics Annual I Vital Statistics and Causes of Death World Health Organization Geneva 1976 130 Mancuso TF Coulter EJ Macdonald EJ Migration and cancer mortality experience - a study of native and sourthern born nonwhite Ohio residents Trace substances in Environmental Health - VI 49-56 1973 131 Schneiderman MA Digestive system cancer among persons subjected to occupational inhalation of asbestos particles a literature review with emphasis on response Env Hlth Persp 307- 311 1974 132. Selikoff IJ Epidemiology of gastrointestinal cancer Hlth Persp 299-305 1974 Env 133 134. 135. 136. Kleinfeld M Biologic response to kind and amount of asbestos J Occup Med 296-300 1973 Blejer mental HP Arlon R carcinogen Talc a possible occupational J Occup Med 92-97 1973 and environ- McMichael AJ Spirtas R Kupper LL mortality within a cohort of rubber 464 1974 An epidemiologic study of workers J 0cc Med 458- Andjelkovic D Taulbee J Symons M Mortality experience of a cohort of rubber workers 1964-1973 J Occ Med 387-394 1976 137. 138 Enterline P Decoufle P Henderson V Mortality in relation to occupational exposure in the asbestos industry J Occup Med 897-903 1972 Elmes PC Wade OC pleural malignancy Relation between exposure to asbestos and in Belfast Ann NY Acad Sci 549-557 1965 JACKSON 025726 APPENDIX I JACKSON 025727 287 Number 001 002 003 004 005 006 007 008 009 0.10 Oll 012 013 014 015 016 017 018 019 020 021 022 023 024 025 026 027 028 029 030 031 032 033 035 036 037 038 039 040 041 042 043 044 045 046 047 048 049 050 051 052 OCCUPATIONAL TITLES -- BPTR BPTR BYTU Batch Preparation Batch Preparation Flaps & Bladders PIEE Pigment Blending CEMG CMTR CMIU SBPR SRPU Cement Mixing Cutting & Milling Cutting & Milling Service Preparation Service Preparation /'Milling ~ MIPL Plystock MTFB Milling & Bladders MITU Milling MLMS CALO CALT SULC PLHA SLIN Miscellaneous Calender Operation Calender Tending Roll Changing Trucking Plystock landling Band Liner Service reroll mend & Service Building clean TUED TUEU TUBC STTU STOX STOD Tuber Operation Tuber Operation Tuber Operation & Bladders Tuber Service Tuber Tuber Service Service & Bladders Tuber incl booking & slitting CEMD BEBI Cementing Read Building TRBL STBE INRE PALT VVMA FUSP Tire Building Service & Bead Building Inspection Repair & Sort Green Paint & Green Tires Tires Valve Preparation Tube & Flap Building incl trucking CKTR Curing CRTU CKFB CRFW Curing Curing Curing black white CREA Bladders CRVV SMOL FIRA FIRE FITU FINX Valves Mold Cleaning & Repair Finishing & Inspecting Repairing Tires Finishing & Repairing & Airbags Finishing & Repairing Bladders & Sleeves MNPT MNWR MNEL MNSM Maintenance Maintenance Maintenance Maintenance Metal MNWL Maintenance NNEC Mechanic MNCP MNMA MNMS Carpentry Machinist Maintenance ; pipefitters cement finisher etc. ae JACKSON 025728 288 Number 053 054 055 056 057 058 059 060 001 062 063 054 055 066 067 068 069 070 071 072 073 074 075 076 077 078 079 080 081_ 082 083 084 085 086 1087 1087 088 089 090 091 092 093 094 095 096 097 098 099 100 101 102 103 104 105 106 107 108 109 Code PHPT TEST TCKR CLRK SJAN STOR OTER MIHC RCPR RCDV RCMT SREC SBRF CMCP KOCP FICP SRCP FOAM ROCK URET PLST RSMF PFMK PEFB PFFN BPMC INDA CRSX INDC OTNT BPMC INDA CRSX INDC REDE OTHR SYNT SYNT SYNT METL MECH SPCP RPFI PLST FUCE ARSP METL MECH SALA UNKN INAC RETI SICK LOFF ACTV DEAD LOAD OT Power Plant Quality Control Testing Trucking genl Clerical Janitor Shipping & Receiving Other Rubber Worker tires & tubes Mill Mixing Prep Reclaim Devulcanizing Reclaim refining Service Service Misc =: Cpd & Chem Prod Reaction Chem Prod Finishing dry -Chem Ship & Chom Prod Manufacture Foam Prod Prod Rocket Liner Urethane Mfg Vinyl Prod Mfg Resin Prod Mfg chemigum Pliofilm Mixing Pliofilm Fabrication pliolite dispersite 345A split Pliofilm Finishing = Batch Prep Prod Prep Prod Curing Pr^d Finish & Inspect Prod tube Other Rubber Worker- & tube Batch Hose & Delt Fabrication & Belt Curing & Belt Finish & Inspect & Bolt Special Rubber Prod Development Other Worker rubber as 381A & police Batch Prep Synthetic Latex - Preparation & Synthetic Latex Finishing & Synthetic Latex Rim Plant Treatment 357A cafeteria etc. Rim Forming & Machining Rim Welding Rim Finishing & Inspecting Production & Manufacture Fuel Cell Aerospace Metal PreparationMechanical Building Salary Unknown Inactive File degreasing operations Retired Sick Suspense Layoff Active Dead Leave of Absence JACKSON 025729 289 Number 110 111 112 13 114 115 116 117 118 119 120 121 1.22 123 124 125 Code MSAB ANOD REPR RETR ADIE EXIT DRET MILT TBCR FOOT RCIP PWSV PWRF PWDA PWMF PWNR ,, Misc Anode Absence Process Coagulation dip Rubber Preparation Retread Adhesives - Exit From Work for Other Reasons = D-isability Retirement Absence for Military Service Tube Building & Curing Footware Industrial Products Prior Work With Solvents or Containing Materials Prior Work With Radio Frequency Energies or Energy Enitring Devices Devices Devices Prior Work Commonly With Known Associated Dermatitis Producing Agents or Environeer.I With Dermatitis Problems Excluding Solvents Prior Work With Metal Fune Prior Work Experience Shows No Potential Exposures Relevant To Thi Study JACKSON 025730 290 CEND CEMG CLRK CMTR CNTU CRRCRRAA Bead Building Batch Preparation Batch Preparation Calender operation Calender tending Cementing Treads Cement Mixing Clerical Cutting & Milling Cutting & Milling Curing bladders bladders bladders involves This involves all proceSSON associated with building beads. Mixing of rubber accelerators antioxidants filters etc. in banburys Includes - and helpers B.B. operators operators Same as 01 except for tubes flaps and bladders This O.T. is for those individuals who operate the calanders This includiasncludianclusdas individuals Firestone 0.T.'s 0.T.'s those on operations as well treads After the treads proper size the cement put on them builders put the tire they will stick After internal mixing in batch is milled into long sheets she ts sheets and then cut into long TheThe stacked or pelletized The pel atspellats pelats sheets are then either returned pelats pellats cr to the BS or milled preparatory to further use calender for plystock tuber for . treads Same as CHTR only only only only the the stock stock is is for for tubes flaps or blbaladdderds ers ea tel a ale aaa TS ~ ek A ia OS lee at ek i nat i a Dea a de) atedTe aR ae DIR RE een ne ce ae et JACKSON 025731 GV Code CRFB CHIN CRSX CRTR CRYV FINX FIRA FIRB FIRB FITU FUSP INKE INKE MECH MIFB MING MIMS. MITO Occupational Tile Curing black Curing white ) Miscellaneous Curing Curing Curing Repairing tires Finishing & Repairing Tubes and Airbags flap Tube Tube and and flap building Inspection & repair - Green tires Mechanical building Milling & Bladders Mill Mixing Milling - Miscellaneous Milling - plystock Milling - treads. 291 DescriptiDeoscrniption Description Mold press cure of flaps Pot heater cure of talc coated flaps Industrial Products Same as FITU except for flaps bladders and sleeves Trim balance label tube & inflate classify service force grind and cured tires Repair buff rag rib repairs of cured tires Final inspection and repair of tubes and airbags Splicing and bookbooking ing tubes & flapflsaps Inspect green tires repair preparatory Largely machinist type work Same as MITD only the slugs for flaps & bladders Similar preparation done on mill by hand to Batch milli.e. onto a pigments This has largely been replaced This is milling that does not fit any where of the above products or where it is . not known what the product is Same as MITD only it is sent to a calender for making plystock chafers inner liner etc. The mixed rubber with all its milled additives is so that the rubber is hot and plastic This tread stock is then sent to a tuber eet rare Ge ONG aad MAUD TATED TE RASTER, Te TR HO NT a OR ETE RMIT EE te Pin! oY NRO IEEE @ WA TM 7 JACKSON 025732 292 MITU MAGP MAGP MNEC MAEL MNMA HNMS MNPT MNSM MANL MNHR OTHR Milling - tubes Mechanic Carpentry Maintenance Machinist Maintenance miscellaneous : Maintenance - painting Maintenance - sheet metal Maintenance - welding Maintenance - millwright Others PALI Paint & line - green tires PIBE Pigment blending PLHA | . Plystock handling PWPT RCDV Power Plant Devulcanizing - reclaim RCMI Milling - reclaim Includes pipefitters coment finishers and other maintenance jobs not included in 44-51 Ventilation ducts ES collectors etc. Machine Repair Jobs not included in above and production guard waitress ~ caferteria helper etc. Spray paint onto green tires before curing Talc is applied to some tires and treads jammed in others Weighing & mixing of pigments accelerators antioxidants for batch building This O.T. includes the splicing of material making plystock for tires and band building All operations associated with power plant Processes of devulcanization heating fiber separation grinding batch preparation digestion batch preparation digestion dewatering drying Processes of mixing & blending refining straining packaging extruding batch building JACKSON 025733 293 RCPR RETR SALA SEPR SJAN SLIN SHOL SRLC STBS STOO Occupational Title Preparation - reclate Retread Salary Service - Batch Preparation - Tires Service - Batch Preparation - Tubes Special Products Janitor Liner Serivce reroll mand clean Mold cleaning & repair Service - reclaim service Roll changing including trucking & Service - tire & bead building building Tuber service - tread tuber including booking & slitting Description Processes of digestion heating cracking shifting sorting agustic separation collection and storage Truck materials to BB for mixing mixing such as Lube stock from CHTR CHTR to through be run supply mixed BS a second time pigments ok Same as SBPR only for tube flap and bladder stock Largely defense products made during WWII Some janitors e.g. 88 cleaners be in service operations up by the and mands plystock liners after the plystock is used tire builders and cleans the liners when necessary Includes trucking of molds to be cleaned All service operation associated with reclaim operation These are service people involved in plystock preparation It includes those changing the rolls on the calenders bias cutting machines machines storagstoerage truckers taking rolls to arees etc. Trucking plystock tread inner etc.and liners builders and materials to tire bead builders Similar stock to STTU except Includes slitting and putting cemented JACKSON 025734 294 STOR STOX STTU Shipping & receiving Tuber service and bladders flaps Tuber service - tubes SYNT TEST Synthetic plant Oo Quality control testing TCKR TEBL TUSC TUBD TruckingTrucking general) . Tire building bladdersoperation Tuber operation - flaps and bladders : Tuber operation - treads TUBU UNKN VVHA Tuber operation - tubes > Unknown _ Valve Preparation All activities of receiving receiving unloading unloading packing and loading materials and finished products Similar to STTU except and bladkars - cutting booker etc. for flaps slugs This would include service operaticsoperatics related to tubers making tubes such punching as cutting tubes punching valves putting on valves Manufacture of synthetic rubber All activities associated with quality control pigments stocks tires tubes etc. All trucking trucking jobs jobs above not included in service Same as TUCD and TUBU except flaps and bladders are made This OT runs the tuber which takes the stock from a mill puts it through a dieforming a tread Similar to TUSU only a tube is formed instead of a tread Do not know enough about to put in above OT's or fit above OT's the jab does not Making spraying and buffing valves JACKSON 025735 295 OCCUPATIONAL TITLE GROUP DICTDICTONAIRY DIOCNTIAONRARYY Batch Preparation Milling Calendering CalendeCrainlgendering Tuber Fabrication and beads tires Tubes Curing preparation Curing Inspection Raw ingredicals rubber fillers oils and are weighed and mixed together in a banbury particulates Environmental exposures to particulates particulates Component Of's BPTR BPTU PIBL CHAG SBPR SBPU MING BRO BPMC RUPR- are CHIR Rubber mix from the banburys is processedfurther and heat state pressure till Component OT's the rubber is in a MITO MIPL MIFB The rubber stock from the mill is rolled out into a sheet of controlled and spliced for use thickness The in building the plystock plystock is cut ply of the tire Some exposure to rereaactcion tion products and solvents exitsts Component OT's CALO CALT CALT SHLCPLHA SLIM Rubber stock from a mill a tread or tube The tread and cemented with solvents tacky surface CEMD Component on the OT's through in cut to TUBO TUBC provide STOX provide STOO wheel Beads are rubber and the rim of the covered wires that rest on in a finished tire Treads beads and plystock are used to build the tire Major exposures are to solvents Component OT's BEBL TRBL STBB The tube from the extruder is cut valves tube and the inside of the spliced together Valve is coated making involves patch and sprayed OT's to the valve The rubber is then cured with a solvent to make it tacky Component TUSH SITU VVMA FUSP CRTU CRVV FITU Spraying green tires or tubes bbeeforfe ore curing repairing and inspectinggreen tires or tubes curing Component OT's IMRE PALI before The green uncured product is curedcured under heat and and pressure Major environmental exposures are products curing fumes talc CREA CRFB CREW CREA CHVV SMOL Component to reaction Of's CRTR The cured tire made ready for FIRA FIRB is inspected repaired shipping finishing if necessary necessary OI's Component and OI's JACKSON 025736 296 10 Maintenance 11 General Service 12 Shipping & receiving 13 ReclaimReclaim 15 Plastics 16 Preparation other than tire and tubes 17 Finaofl inspection tubes 18 Metal handling 19 Other rubber This OIG includes welders painters Component OT's MRCP pipefitters metal, pipefitters electy workers workers workes and MAIL This is a miscellaneous group with a jobs and exposures exposures exposures Component Of's SJAN OTHR OTHR OTHP variety unloading carbon Receiving and storing all incoming The that go into making making the final product e.g. carbon pigments new rubber etc. Loading the final product trucks and boxcars Component CT's STOR Scrapped products Scrapped vulcanized rubber are shredded up and devulcanizeddevulcanized The rubber compound is restored restored its original plastic state through application of heat chemicals Primary exposures are to particulates and devulcanizing agents Component OT's RCPR RODY RCMI SREC Chemical products SYNT plant are compounded OTG Component OT's PFMX PEFBPEFB PFFN SYNT Jobs in this group included the manufacturing of urethane and vinyl products with a variety of exposures Component OT's FOAM URET PLST Raw ingrediantsare banbury milling Component OT's IMDA The cured tube made ready for INDC is inspected repaired shipping finishing if necessar and Component CT's being currently Rim plant jobs are minimum A minimum amount and metal handling welding offer offer Akron phased out in type jobs involving but very little little exposure to lubricating oils solvents and and particuletepartisculetes Component OT's METL MECH SPCP PRPT This OTG includes a variety of different jobs shore products other than tires and tubes are produced produced fooluar fooluar adhesives and fuel cells for example Component OT's ROCK OTNT REDE FUCE ARSP ANOD ADHE FOOT JACKSON 025737 297 016 20 Satered 21 Unknown GENERAL DESCRIPTION DESCRIPTION DESCRIPTION DESCRIPTION This 916 includes managerial secretarial and other office jobs Foresten ere not included here rather are assigned to the other 010's Component of's SALA Not enough information on the work history to assign to specific 01G Component Of's UNKN JACKSON 025738 298 SUMMARY OF MAJOR POTENTIAL EXPOSURES TO CHEMICAL AGENTS IN TIRE MANUFACTURING OTG's Compounding and Mixing Accelerators e.g. thiazoles guanidines thiruam sulfides carbamates Antioxidants Fillers carbon black Extender Cutting and Milling Milling Extrusion Talc . Emission from Vulcanization Process Solvents Calendering Tire Building Emissions from Vulcanization Process Solvents Curing Preparation Curing Final Inspection Variable exposure to solvents talc _ Emission from Vulcanization Products Cured rubber dust solvents Emissions from the hot rubber mix can be of two sorts based on theoretical considerations and will not be the same for milling calendering and curing 1 Volatization - a Residual monomers probably only milling and calendering styrene acrylonitrile chloroprene b Antioxidants and Antiozonants and Accelerators - higher loss in milling and calendeinrg C. Processing Aids - oils d Miscellaneous Ingredients - special ingredients amines esters organic acids and impurities 5-40 of main ingredient 2 Formation of New Compounds - Oxidative products are formed in milling and calendering reductive products are formed in the mold during curing oxidative products when the curing press is opened The compound classes formed include amines and ammonia organic sulfides hydrocarbons acids esters water It has been estimated that 0.5-0.7 volatile material would be released and conceivably 1,000 or more different compounds JACKSON 025739 APPENDIX II JACKSON 025740 . APPENDIX II 300 PG2 | OHSG COMPUTER GROUP UPDATE O MAY 1977 VARIABLE CONTROL STUDY PROGRAM OCCUPATIONAL HEALTH STUDIES GROUP SCHOOL OF PUBLIC HEALTH UNIVERSITY OF NORTH CAROLINA CHAPEL HILL NORTH CAROLINA 27514 JACKSON 025741 1.0 GENERAL DESCRIPTION 301 1.1 ABSTRACT The Variable Control Study program is designed to produce statistics for control studies when either a fixed or a variable number of controls are matched to the cases in file The statistics may be generated for one or more fields in the records and they are 1. odds ratio 2. Chisq 3. Chi 4. Confidence Interval 1.2 SOURCE Allan Smith Department of Epidemiology requested the development of this program supplied the formulas and also participated in the documentation It is an expansion of an earlier OHSG program designed by Ravi Mathew Programmer and Professor Anthony McMichael Depart- ment of Epidemiology which analyzed control files that had a fixed number of controls matched to a case 1.3 COMPUTATIONAL PROCEDURE For control studies a case record and the one or more control records that are matched to that case make up a single entity for analysis purposes For these groups formulas have been developed which uses a by matrix as the basic component for solving the equations The matrix contains exposure information on a field to be analyzed JACKSON 025742 by Matrix Case Control Exposed A Ci Exposed B. Wi | D. X. U V. Y. where i = a group which consists of a single case and the controls that are matched to that case A. 1 = if the case is exposed and 0 if unexposed B. ====== 1 if the case is not exposed and 0 if exposed C = The number of controls exposed Di = The number of controls not exposed U = The sum of A + C V = The sum of B. + D. W. = The sum of A + B Xi = The sum of C + Di Y The sum of U. + V. or the sum of Wi + X 302 Formulas based on the by matrix Odds Ratio === A D. i Ys B5 B5 B5 B5 JACKSON 025743 303 Chi 11 \ Chisq Confidence where Z Interval = Exp value for normal in odd of 1 - z value value distribution is 1.96 for 95 confidence limit and 1.645 for a 90 confidence limit There can be more than one field in the Data records for which the analysis is desired and the concept of the by matrix would have to be produced for each of them The program will treat all blanks or zero as not exposed and treat any other value in a field as exposed Changing of values meet these criteria must take place outside of this program if necessary 2.0 INSTRUCTIONS FOR USE 2.1 ORDER OF JOB DECK JOBNAME JOB UNC.OHS ..... aceeueee RE OFS JOBT CARD .........---. PASSWORD ecgreeeees // EXEC CASE STEPLIB DD UNC.OUHNC.OSHS.F.207F.FLOAD2UNC.O0HS.F207F.7LOAD UFNC.OHS..F207FL.LOADOUNC.OAHS.F207DF.LOAD * CONTROL DD CONTROL INFORMATION DATA DD DISP . _ eevee .DEFINITION.OF.DATA.FILE .DEFINITION.OF.DATA.FILE SYSPRINT DD SYSOUT . JACKSON 025744 304 2.2 PREPARATION OF CONTROL CARDS Card 1 All of the fields on this card must be numeric each field must be separated by one or more blanks and the values must be contained within column 1-79 1. The length of a DATA record 2. Location of an ID field in the DATA record which identifies case a case and the controls that are matched with that 3. Length of the ID field 4. Location of a control in the DATA record which identifies an individual record as being a case record or a control record 5. Length of the control number . 6. The value of the control which identifies a . | a case 7. Maximum number of control records that are matched to any one case 8. Location of the first field in the DATA record for which an exposure check is to be made 9. Length of a field 10. Number of digits in a field which denotes a 11 12. fraction The number of fields in a record The value which is to be used in the confidence interval formula JACKSON 025745 305 Card 2 This card contains a title which will appear on the print out above the tables generated it may be alphanumeric contained in column 1-80 and column 1 must be a blank character Card N These cards contain labels which will appear on the print out identifying the fields Their use is optional If they are not given the program will label each field of as OT 1 - OT N where N is the number fields in a record as will If they are given there may be as many put on a card fit there must be one for each field they may be alphanumeric and must have a length of 1-8 characters must be enclosed within quote marks and each name with its enclosing quote marks must be separated by one or more blanks 2.3 DATA FILE The data file must be sorted in ascending order by ID and casecontrol order and there must be a separate record for each case and control as follows First case record Control 1 for the first case record Ld Control N for the first case record Second case record Control 1 for the second case record JACKSON 025746 306 The required fields in a record are 1. An ID field which identifies a case and the controls which are matched to that case This field may be alphanumeric and may contain blanks 2. A control which identifies a record as being a case or as being a control This number must be all numeric and cannot contain leading blanks even if more than one digit long . 3 The fields These fields may contain blanks numbers and a decimal point If a decimal point is included in the value it overrides the fractional specification given in the Control cards If a decimal point is not included in the value the fractional part is taken to be the number of rightmost digits specified in the Control cards All of the field which are to be analyzed must be adjacent to one another In order for a group of records having the same ID to be included in the analysis they must pass the criteria of the above items as well as the following " There must be a case record and some control records for each ID 2 There cannot be duplicate case numbers or duplicate control numbers for an ID 3 The value that identifies a record as being a control must be greater than the value that identifies a record as a case 4 The number of control records matched with any one case cannot exceed the maximum specified in the control cards JACKSON 025747 2.4 EXAMPLE EXECUTION B109 JOB UNC.OHS.F299E LEGGETT 9907 // EXEC CASE STEPLIB DD DISP DSN UNC.OHS.F207F.LOAD UNC.OHS.F207F.LOAD UNC.OHS.F207F.LOAD CONTROL DD * 163 5 2 4 1 1 4 23 73 73 B109 Matched analysis on Memphis 1975 data TBLD CADM QRSM TTLD TTLD DATA DD UNC.OHS.F9907.MEMUNCP.OHS.HF990I7.MEMSPHIS SYSPRINT DD SYSOUT 307 5 2.24 3.0 EXECUTION TIMING The run time used to process 616 records with 21 exposure fields to analysis on each record was 10.8 seconds on an IBM 370/165 computer The core storage used was 102K JACKSON 025748 APPENDIX III JACKSON 025749 309 OBSERVER AGREEMENT ON ESTIMATE OF POLYCYCLIC HYDROCARBONS LEVELS OCCUPATIONAL TITLE : High | Med | Low None {| Unk | 001 Batch Preparation 1 _ 002 Batch Preparation Flaps Bladders 111 1 003 Pigment Blending 004 Cement Mixing 1 005 Cutting & Milling 11 006 Cutting & Milling 007 Service Preparation 008 Service Preparation 009 Milling 010 Plystock 011 Milling & Bladders 012 Milling 013 Miscellaneous 014 Calender Operation 015 016 Calender Tending Roll Changing Trucking & Service 017 018 Plystock Handling Band Building Liner Service reroll mend clean 019 Tuber Operation - 020 Tuber Operation 021 Tuber Operation & Bladders 022 023 Tuber Service Tuber Service & Bladders 024 includslitatnd i slinttigng 1 1 111 111 2 1 2 1 1 11 1 1 2 1 2 1 2 1 2 1 2 1 2 1 2 2 1 2 1 1 2 712 712 12 12 2 2 - 2 1 2 1 -2 1 2 JACKSON 025750 310 OCCUPATIONAL TITLE 025 __ Cementing 026 Bead Building High Med | Low 2 2 None | Unk 1 I 027 Tire Building 2 028 029 030 Service & Bead Building & Sort Green Tires Inspection Inspection Inspection Inspection Repair Repair Paint & Green Tires 2 1 1 2 3 031 Valve Preparation 032 Tube & Flap Building 033 Curing 1 2 1 2 1 1 1 034 Curing 035 Curing black 036 Curing white * 037 Bladders 038 Valves 039 Mold Cleaning & Repair 040 041 042 043 Finishing & Inspecting Tires . Repairing Tires Finishing & Repairing & Airbags Bladders Finishing & Repairing 044 Maintenance 045 Maintenance 046 Maintenance 047 Maintenance Metal 048 Maintenance 1 1 1 1 1 1 1 1 1 I 1 1 1 2 1 14 2 21 2 1 - 1 o> 2 2 21 1 2 1 2 1 2. 1 21 21 1 1 JACKSON 025751 311 . 049 OCCUPATIONAL TITLE Mechanic 050 Carpentry 051 Machinist High | Med | Low | None | Unk 21 21 2 1 21 21 053 054 Power Plant Quality Control Testing 055 Trucking genl 056 Clerical 057 Janitor 058 059 Shipping & Receiving Other Rubber Worker tires & tubes 060 Mill Mixing 061 Prep - Reclaim 062 Devulcanizing - Reclaim 063 Milling - Reclaim refining 064 Service - Reclaim 065 Service - BB - Miscellaneous 066 Compounding & Mixing - Chem Prod 067 Reaction Operator- Chem Prod 068 Finishing dry bag - Chem Prod 069 070 Shipping & Receiving - Chem Prod Manufacture Foam Prod 071 Rocket Liner Mfg 072 Urethane 1 12 21 1 1 2 111 21 21 111 111 1 1 12 12 1 1 | 12 12 111 12 21 1 1 3 1 1 1 111 111 111 111 1 111 21 21 2 1 21 21 2 1 2 1 3 JACKSON 025752 312 - OCCUPATIONAL TITLE 073 074 075 Vinyl Prod Mfg Resin Prod dispersite Mfg 345A scplhietmigum , pliolite pliolite , Pliofilm Mixing 076 Pliofilm Fabrication 077 078 Pliofilm Finishing Batch Prep - Ind Prod 079 Prep - Ind Prod High Med { Low | None ] Unk 3 3 3 fl | . {| 3 111 111 111 2 H 080 Curing - Ind Prod 1 1 081 Finish & Inspect - Ind Prod 082 - Other Rubber Worker - tire tube 083 Batch Prep - Hose & Belt 4 1 084 Fabrication ... Hose & Belt 2 085 } 086 Curing - Hose & Belt Finish & Inspect Ab Hose & Belt 1 1 087 Special Rubber Prod Development 088 Other Worker rubber such as police cafeteria etc. 089 Batch Prep - Synthetic Latex 090 Preparation & Synthetic Latex . 091 Finishing & Synthetic Latex 092 Rim Plant 1 Metal Treatment 1 2 1 2 1 1 1 1 2 l 2 2 2 2 2 3 093 094 Rim Plant - Forming & Machining | Rim Welding 095 Rim Finishing & Inspecting 096 Production & Manufacture "' | | 3 3 a 3 . 3 JACKSON 025753 313 OCCUPATIONAL TITLE 097 Fuel Cell 098 099 100 Aerospace Metal Preparation mostly degreasing oper Mechanical Building 101 Salary 111 Anode Process Coagulation dip 112 Rubber Preparation 113 Retread 1.14 Adhesives 118 Tube Building& Curing 119 Footware High _ 1 Low None | Unk 2 1 2 3 2 1 3 2 2 3 fl 2 3 JACKSON 025754 314 OBSERVER AGREEMENT ON ESTIMATE OF ASBESTOS LEVELS OCCUPATIONAL TITLE High } Med [ Low | 001 002 Batch Preparation _ Batch Preparation Flaps Bladders 003 Pigment Blending 004 Cement Mixing 005 Cutting & Milling 006 Cutting & Milling 111 007 Service Preparation 111 008 Service Preparation 009 Milling 010 Plystock 011 Milling & Bladders 012 Milling 013 Miscellaneous 014 Calender Operation 015 Calender Tending 016 Roll Changing Trucking & Service 017 018 Plystock Handling Band Building Liner Service reroll mend clean 019 Tuber Operation 1 020 Tuber Operation 021 Tuber Operation & Bladders 022 Tuber Service 023 Tuber Service & Bladders 024 includ includ booking andslitting 12 12 1 2 412 412 12 1 111 111 111 111 12 12 1 2 21 2 2 3 1 2 3 12 12 HB None | Unk 12 21 21 1 1 1 3 3 3 3 q 3 3 JACKSON 025755 315 OCCUPATIONAL TITLE 025 Cementing | 026 Bead Building ; 027 Tire Building 028 029 030 031 Service & Bead Building Inspection Repair & Sort Green Tires Paint & Green Tires | Valve Preparation 032 Tube & Flap Building 033 Curing 034 Curing 035 Curing black 036 Curing white 037 Bladders 038 039 Curing Mold Cleaning & Repair 040 041 Finishing & Inspecting Tires . Repairing Tires 042 043 Finishing & Repairing & Airbags Finishing & Repairing Bladders 044 Maintenance 045 Maintenance 046 Maintenance 047 Maintenance Metal 048 Maintenance High | Med f[ Low . 2 2 2 I 3 1 - 1 2 1 21 21 21 111 None Unk | 3 3 3 3 1 1 2 1 1 2 2 3 3 21 21 21 21 21 111 111 21 21 2 1 JACKSON 025756 316 OCCUPATIONAL TITLE | Mechanic | 049 050 Carpentry 051 Machinist ~~ High | Med | Low 4 111 | None | Unk 2 L 1 l | 111 111 . 053 Power Plant 054 Quality Control Testing 055 Trucking genl 056 Clerical 057 Janitor 058 059 Shipping & Receiving Other Rubber Worker tires & tubes | 060 Mill Mixing 061 Prep - Reclaim 062 063 Devulcanizing - Reclaim Milling - Reclaim refining 064 Service - Reclaim 065 Service - BB - Miscellaneous 066 & Mixing - Chem Prod Compounding 067 | 068 | 069 | 070 - Chem Prod Reaction Operator Finishing dry bag - Chem Prod & Receiving - Chem Prod Shipping Manufacture Foam Prod 071 Rocket Liner Mfg 072 |__ _ Urethane . 4 1 1 H 2 1 1 2 1 2 2 2 11 11 1 1 3 I 2 1 3 3 2 1 1 2 2 2 , 2 ---- 21 1 1 l 1 21 3 JACKSON 025757 317 - OCCUPATIONAL TITLE 073 Vinyl Prod Mfg | High Med ; Low | None | Unk 2 1 075 Pliofilm Mixing - 076 Pliofilm Fabrication 077 Pliofilm Finishing 078 Prod Batch Prep - Ind . 079 Prep~ Ind Prod 080 - Curing Ind Prod 081 Finish & Inspect - Ind Prod 082 Other Rubber Worker - tire tube 083 Batch Prep - Hose & Belt | 084 Fabrication - Hose & Belt | 085 Curing - Hose & Belt 086 Belt Finish & Inspect - Hose & 087 088 = 089 Prod Development Special Rubber Other Worker rubber such as police cafeteria etc. Batch Prep - Synthetic Latex | 090 Preparation & Synthetic Latex 091 Finishing & Synthetic Latex 092 Rim Plant - Metal Treatment 093 Rim Plant - Forming & Machining , | 094 Rim Welding 095 5 096 Rim Finishing & Inspecting & Manufacture Production - | ~ 2 2 1 3 1 2 . 3 ' 1 1 2 3 2 3 3 2 3 3 3 3 3 3 | | 3 ; : 3 1 _ 1 JACKSON 025758 318 OCCUPATIONAL TITLE 097 Fuel Cell 098 Aerospace 099 mostly degreasing oper 100 Mechanical Building 101 Salary 111 Anode Process Coagulation dip 112 Rubber Preparation 113 Retread 114 Adhesives 118 Tube Building& Curing 119 Footware High Med Low | None | Unk 2 1 2 I ; 3 2 3 3 2 I 3 3 2 1 | 2 X JACKSON 025759 - 319 OBSERVER AGREEMENT ON ESTIMATE OF NITROSAMINE LEVELS OCCUPATIONAL TITLE - High | Med Low None | Unk 1 001 Batch Preparation 002 Batch Preparation Flaps Bladders 111 3 003 Pigment Blending 004 Cement Mixing 005 006 Cutting& Milling Cutting & Milling 007 Service Preparation 008 Service Preparation 009 Milling 1 010 Plystock 11 011 Milling & Bladders 1 11 012 Milling 1 013 Miscellaneous 014 Calender Operation 015 Calender Tending 016 Roll Changing Trucking & Service 017 Plyscock Handling Band Building 018 Liner Service reroll mend clean 0.19 Tuber Operation 020 Tuber Operation 021 Tuber Operation & Bladders 022 Tuber Service 023 Tuber Service & Bladders 024 includ Service Tread Tuber 1 1 111 111 1 1 1 1 2 1 2 1 11 1 11 1 111 111 111 111 A 21 3 1 1 1 1 1 1 1 111 111 2 21 12 12 7111 7111 7111 111 111 111 1 11 11 12 12 712 712 1| 2 JACKSON 025760 320 a OCCUPATIONAL TITLE 025 026 Cementing Treads Bead Building 027 Tire Building 028 029 030 Service & Bead Building Inspection includ trucking trucking trucking & Sort Green Tires Paint & Green Tires 031 Valve Preparation 032 Tube & Flap Building 033 Curing 034 Curing 035 Curing black 036 Curing white 037 Bladders 038 i 039 Curing Mold Cleaning & Repair 040 Finishing & Inspecting Tires 041 042 043 Repairing Tires Finishing & Repairing & Airbags Finishing & Repairing Bladders 044 Maintenance - 045 Maintenance 046 5 047 Maintenance Maintenance Metal 048 Maintenance High Med Low None Unk I 2 1 2 1 2 | 3 3 | 3 3 1 2 1 1 H - 1 1 1 i 1 2 1 1 1 1 1 11 11 1 2 1 2 1 32 21 21 21 21 21 21 2 1 111 111 111 21 21 I 1 1 JACKSON 025761 321 OCCUPATIONAL TITLE 049 Mechanic 050 Carpentry . | 051 Machinist . | High | Med | Low | Nona | Unk 21 21 2 1 2. 1 053 Power Plant . 054 Quality Control Testing 055 Trucking genl 056 Clerical 057 Janitor 058 Shipping & Receiving 059 Other Rubber Worker tires & tubes 060 Mill Mixing 061 Prep - Reclaim } 062 Devulcanizing - Reclaim 063 064 Milling - Reclaim refining Service - Reclaim 065 Service BB - Miscellaneous 066 Compounding & Mixing - Chem Prod 067 Reaction Operator - Chem Prod 068 Finishing dry bag - Chem Prod 069 Shipping & Receiving -- Chem Prod 070 Manufacture Foam Prod . 071 Rocket Liner Mfg 072 Urethane | 21 1 11 . : 2 1 1 1 1 21 12 111 111 111 111 1 3 2 1 111 11 12 | 1 2 1 1 1 1 3 1 1 2 1 1 12 111 111 N 1 1 1 1 111 111 1 12 2 JACKSON 025762 322 41 OCCUPATIONAL TITLE 073 Vinyl Prod Mfg High | Med | Low None | Unk 2 1 | 075 Pliofilm Mixing 076 Pliofilm Fabrication 077 078 Pliofilm Finishing Batch Prep - Ind Prod 079 Prep - Ind Prod 080 Curing - Ind Prod 2 H 2 1 1 081 Finish & Inspect -- Ind Prod - OS2 Other Rubber Worker tire tube 083 Batch Prep - Hose & Belt 2 1 084 035 Fabrication - Hose & Belt Curing - Hose & Belt 1 1 086 Finish & Inspect - Hose & Belt Prod Development 1 087 Special Rubber 09098 8 Other Worker rubber such as police cafeteria etc. 1 089 Batch Prep - Synthetic Latex 090 Preparation & Synthetic Latex I 091 092 Finishing & Synthetic Latex Rim Plant - Metal Treatment 093 Rim Plant - Forming & Machining 094 Rim Welding 095 096 Rim Finishing & Inspecting ~ Production & Manufacture 21 21 21 21 . 2 1 1 1 2 1 1 1 1 2 1 1 2 1 2 3 2 2 12 12 3 3 3 3 1 1 1 JACKSON 025763 po . OCCUPATIONAL TITLE 097 Fuel Cell 098 Aerospace = 099 Metal Preparation _ mostly degreasing oper 100 Mechanical Building 101 Salary 111 112 Anode Process Coagulation dip Rubber Preparation 113 Retread 114 Adhesives 118 Tube Building& Curing 119 Footware 323 None | Unk 2 1 2 1 ft4 3. 3 3 3 1 1 1 2 1 1 . . JACKSON 025764 324 OBSERVER AGREEMENT ON ESTIMATION OF CARBON BLACK LEVELS OCCUPATIONAL TITLE - High Med Low None | Uak 001 Batch Preparation 3 002 Batch Preparation Flaps Bladders 3 003 Pigment Blending 1 2 004 Cement Mixing 005 Cutting & Milling 006 Cutting & Milling 007 Service Preparation 3 1 1| 2 1 I 1 1 2 008 Service Preparation 1 2 009 Milling 1 2 010 Plystock~ 011 Milling & Bladders 012 Milling 013 Miscellaneous 014 Calender Operation 015 Calender Tending 016 Roll Changing Trucking & Service 017 Plystock Handling Band Building 018 Liner Service reroll mend clean 019 Tuber Operation 020 Tuber Operation 021 Tuber Operation & Bladders 022 023 Tuber Service Tuber Service & Bladders 024 Tuber Service Service TreadTubetting 1 2 1 2 1 2 1 2 3 3 3 3 3 3 3 3 3 3 JACKSON 025765 325 - OCCUPATIONAL TITLE | 025 Cementing Treads | 026 Bead Building 027 028 029 030 Tire Building Service & Bead Building Inspection Truepair & Sort Green Tires Paint & Green Tires 031 Valve Preparation 032 Tube & Flap Building 033 Curing 034 Curing 035 Curing black 036 Curing white 037 Bladders 038 Valves 039 Mold Cleaning & Repair 040 041 Finishing & Inspecting Tires . Repairing Tires 042 Finishing & Repairing & Airbags 043 Finishing& Repairing Bladders 044 Maintenance 045 Maintenance 046 Maintenance 047 Maintenance Metal 048 Maintenance High Med Low | None Unk 3 3 3 3 : f- 3 3 3 1 2 I 2 1 2 1| 2 1 2 . 2 1 2 1 2 1 2 1 1 1 2 2 i 1 111 111 I 1 1 1 1 111 111 1 1 JACKSON 025766 326 OCCUPATIONAL TITLE | 049 Mechanic 050 Carpentry 051 Machinist High Med - Iow | None ] Unk - 1 1 1 2 1 21 21 | 053 Power Plant 054 Quality Control Testing 055 Trucking genl 056 Clerical -057 Janitor 058 059 Shipping & Receiving Other Rubber Worker tires & tubes 060 Mill Mixing 061 Prep - Reclaim 062 063 Devulcanizing - Reclaim Milling - Reclaim refining 064 Service - Reclaim 065 Service - BB - Miscellaneous 066 Compounding & Mixing Chem Prod 067 068 Reaction Operator - Chem Prod Finishing dry bag - Chem Prod 069 Shipping & Receiving - Chem Prod 070 Manufacture Foam Prod 071 Rocket Liner Mfg 072 Urethane 12 H : 12 1 1 1 21 1 1 1 1 1 12 3 12 1 1 " 1 21 2 1 11 11 2 2 2 2 2 3 3 3 ------ 3 JACKSON 025767 327 a = 7 - OCCUPATIONAL TITLE 073 Vinyl Prod Mfg ee - High { Med 7 . [ Low | _ ae None Unk fo 3 075 Pliofilm Mixing 076 Pliofilm Fabrication . 077 Pliofilm Finishing | Prod 078 Batch Prep - Ind 2 . 079 Prep - Ind Prod 080 Curing - Ind Prod 081 Finish & Inspect - Ind Prod 082 - Other Rubber Worker - tire& tube 083 Batch Prep - Hose & Belt 2 084 085 Fabrication A Hose & Belt Curing - Hose & Belt 086 Finish & Inspect - Hose & Belt 087 088 089 Special Rubber Prod Development Other Worker rubber such as police cafeteria etc. Batch Prep - Synthetic Latex | 1 | 3 7 3 3 1 1 1 1 2 . 1 1 2 - 2 1 1 1 . 1 1 2 1 2 1 1 3 3 090 Preparation & Synthetic Latex 3 091 Finishing & Synthetic Latex 3 092 Rim Plant - Metal Treatment 3 093 Rim Plant -- Forming & Machining 3 094 Rim Welding 095 Rim Finishing & Inspecting 096 Production & Manufacture | . ; | ; - 3 JACKSON 025768 328 OCCUPATIONAL TITLE 097 Fuel Cell 098 Aerospace | 099 Metal Preparation mostly degreasing oper 100 Mechanical Building _ 101 Salary 111 Anode Process Coagulation dip 112 Rubber Preparation 113 Retread 114 Adhesives 118 Tube Building & Curing 119 Footware High Med . Low | None | Unk - 7 2 I 2 1 3 2 1 3 3 1 1 1 1 I 1 3 1 1 2 1 1 _~ JACKSON 025769 BIBLIOGRAPHY JACKSON 025770 330 BIBLIOGRAPHY - Aird I Bentall HH Roberts JAF A relationship between cancer of the stomach and ABO blood groups Br Med J 799-801 1953 Andjelkovich D Taulbee J Symons M Mortality experience of a cohort of rubber workers 1964-1973 J Occ Med 387-394 1976 Archer VE Magnuson HJ Holaday DA Lawrence PA Hazards to health in uranium mining and milling JOM 55-60 1962 histologic of Saccomanno G Jones JH Frequency of different bronchogenic carcinoma as related to radiation exposure Cancer 2056-60 1974 types Wagoner JK Lundin FE Lung cancer among uranium miners in the United States Hith Phys 351-71 1973 Armijo R Coulson AH Epidemiology of stomach cancer in Chile - the role of nitrogen fertilizers Int J Epid 301-09 1975 Bera JW Histologic aspects of the relation between gastric adenomatous polyps and gastric cancer Cancer 1149-55 1958 groups Billington BP Gastric cancer - relation between ABO blood and epidemiology Lancet 859-62 1956 , site Bjelke E Epidemiologic studies of cancer of the stomach rectum with special emphasis on the role of diet Gastroen Vol 19 Suppl 31 1974 colon and Scand J Talc Blejer HP Arlon R a possible occupational and environmental carcinogen J Occup Med 92-97 92-97 1973 Boyd JT Doll R Gastrointestinal cancer and the use of liquid pa,, rafin Brit J Cancer 231-37 231-37 1954 . Braun DC Truan TD An epidemiologic study of lung cancer in asbestos miners Arch Ind Hlth 17 634-52 1958 Buckwalter JA Wohlwend CB Colter DC Tidrick RT Knowler LA The 1957 gastric association of the ABO blood groups to Gynecol Obstet 104 176-79 carcinoma Surg JACKSON 025771 331 Buell P Dunn JE Jr of California Cancer mortality among Japanese Cancer 656-64 1965 Issei and Nisei ' Dunn JE Jr Breslow L The occupational class risks of cancer mortality in men J Chron Dis 600-21 1961 Choi NW in Ethnic distribution of cancer of the gastrointestinal Manitoba Amer J Pub Hlth 2067-81 1968 tract Cohart E Socioeconomic distribution of stomach cancer in New Haven Cancer 455-61 1954. Cohn I Jr Sartin J Sudduth P Giant ulcers of the stomach Gastroent 121-35 1959 Amer J Comfort MW Kelsey MP Berkson J development of carcinoma of 367-73 1947 Gastric acidity the stomach J before and Nat Cancer after Inst the Correa P Haenszel W Cuello C Tannenbaum S Archer M A model for gastric cancer epidemiology Lancet 1975 , Crane PS Rhec SV Scel DJ Experience with 1079 cases of cancer of the 4 stomach seen in Korea from 1962 to 1968. 1970 Amer J Surg 747-51 Creagan ET Hoover RN Fraumeni JF Jr Mortality from stomach cancer in coal mining regions Arch Environ Hith 28-30 1974 Crumb CK Willets PF Stephenson HE Gastric cancer - descrescent malignancy Surgery 277-82 1970 Davies RI Wynne Griffith G Cancer and soils in the county of Anglesey Brit J Cancer 56-66 1954 Doll R Mortality from lung cancer in asbestos workers 81-86 1955 Br J Ind Med Doll R The geographical distribution of cancer 1-8 1969 Brit J Cancer Donaldson AW Epidemiology of lung cancer among uranium miners 563-69 1969 . Hlth Phys Dunham LJ Bajlar JC III World maps of cancer mortality rates and frequency ratios J Nat Cancer Inst 155-203 1968 Elliot CV Wald SM Benz RI A roentgenologic study of ulcerating lesions of the stomach Amer J Roentgen 612-22 1957 JACKSON 025772 332 End Results in Cancer Report No. 4. U. S. Dept of Health Education and Welfare National Cancer Institute 1972 Enterline PE Mortality among asbestos products workers in the United States Ann N. Y. Acad Sci 156-65 1965 Flamant R Laserre 0 Lazar P Lequerinais J Denoix P Schwartz D Differences in sex ratio according to cancer site and possible relationship with use of tobacco and alcohol review of 65,000 : cases J Nat Cancer Inst 1309-16 1964 Graham S Levin M Lilienfeld AM cancer of various siteisn 180-91 1960 The socioeconomic distribution of Buffalo N. Y. 1948-52 - Cancer 13 , Levin ML Lilienfeld AM Sheehe P Ethnic derivations as related to cancer at several sites Cancer 13-27 1963 , Lilienfeld AM Tidings JE Dietary and purgation factors in the epidemiology of gastric cancer Cancer 2224-34 1967 Schotz W Martino P Alimentary factors in the epidemiology of gastric cancer Cancer 927-38 1972 Gregor 0 Toman R Prusova F mortality to mortality ; 1974 Relation of in general gastrointestinal cancer Scand J Gastroent Suppl 79-85 Griffith GW The sex ratio in gastric cancer and hypothetical relative to etiology Brit J Cancer 163-72 1968 considerations | Haenszel W Cancer mortality among foreign born in the United States Nat Cancer Inst 37-132 1961 Variations in incidence particular reference to the 213-62 1958 of mortality from United States J stomach cancer Nat Cancer Inst with Hammond EC Smoking in relation to mortality and morbidity first thirty months of follow in a prospective in 1959. J Nat Cancer Inst 1161-88 1964 Findings in study started Henderson WJ Evans DMD Davies JD Griffiths K Analysis of particles in stomach tumors from Japanese males Env Res 240-49 Hill MJ Hawksworth G Tattersall G Bacteria nitrosamines and cancer of the stomach BrJ Cancer 562-67 1973 Hirohata T Kuratsune M The geographical comparison of mortality from cancer of the stomach and ulcer of the stomach in Japan Br J Cancer 465-79 1969 JACKSON 025773 333 aA . Hoover R Mason TJ McKay FW Fravmeni JF Cancer by county new resource for etiologic clues Science 1005-07 1975 Hopps HC Principles of Pathology 1964 Appleton New York Kaplan HS Rigler LG Pernicious anemia and carcinoma of the stomach Am I Med Sci 339-48 1945 Kellerman G Shaw CR Kellermann M Aryl hydrocarbon hydroxylase inducibility and bronchogenic carcinoma NEJM 934-36 1973 Kleinfeld M Biologic response to kind and amount of asbestos Med 296-300 1973 J Occup Messite J Zaki MH Mortality experience among talc workers a follow study J Occ Med 345-49 1974 Kraus AS Levin ML GerhardPtR A study of occupational associations with gastric cancer Am J Pub H1th 961-70 1957 e Lauren P The two histologic main types of gastric carcinoma diffuse and so called intestinal type carcinoma Acta Path Microbiol Scand e 31-49 1965 Lerer TJ Gustin J Kamon E term mortality study among steelworkers VIII Mortality patterns of open hearth steelworkers a preliminary report JOM40-43 1975 , Redmond CK Breslin PP Study among steelworkers operators JOM 608-14 Slavin L Rush HW Long mortality VII Mortality patterns among crane 1974 Levin ML Haenszel W. Carrol BE Gerhardt PR Hardy Cancer incidence in urban and rural areas of Nat Cancer Inst 1243-57 1960 VH New Ingraham SC York State III J Lloyd JW Ciocco A term mortality study of steelworkers I. Methodology JOM299-310 1969 Lloyd JW term mortality study of steelworkers V. cancer in coke plant workers JOM53-68 1971 Respiratory > Lundin FE Jr. Redmond CK Geiser PB Long mortality study of steelworkers IV Mortality by work areas JOM 151-57 1970 Lombard HL Doering CR Cancer studies Massachusetts III Cancer Mortality in nativity groups J Prev Med 343-61 1929 JACKSON 025774 334 Lorenz - E Radioactivity and lung cancer canceirn miners of Schneeberg and Inst 1-15 1944 a critical review of lung Joachimsthal ] Nat Cancer Low H Nitroso compounds - safety and public health 256-60 1974 Arch Environ Hith Lundin FE Wagoner JK Archer UE Radon daughter exposure and respiratory cancer - quantitative and temporal aspects S. Dept of HEW PHS NIOSH and NIEHS Joint Mangr No. 1 1976 Lynch KM Smith WA Pulmonary asbestosis III - carcinoma of lung in asbestosis Amer J Cancer 56-64 1935 = MacDonald WC Clinical and pathological features of adenocarcinoma of gastric cardia Cancer 724-32 1972 the Magee PN Barnes JM in the rat by 1956 The production of malignant feeding dimethylnitrosamine primary hepatic tumors Br J Cancer 114-22 Magee PN Possibilities of hazard from nitrosamines in industry Occ Hyg 19-22 1972 Ann Mancuso TF Coulter EJ MacDonald EJ experience - a study of natives residents Trace substances in 1973 Migration and cancer mortality and southern born white Ohio Environmental Health - 49-56 Migration and and southern born 49-56 1973 cancer mortality experience - a nonwhite Ohio residents Trace study of native Subs in Env Hlth Marshak HH Feldman F Gastric polyps Am I Dig Dis 909-35 1965 Matolo NM Gorishek WM Moslander V Dixon JA Coal mining and cancer of the stomach Rocky Mount Med J 44-49 1972 Mazumdar S Lerer T Redmond CK term mortality study among workers IX Mortality patterns among sheet and tin mill JOM751-55 1975 steel- workers McDonald JC McDonald AD Gibbs GW Mortality in the chrysolite asbestos ' mines and mills Quebec ArEnc v Hh lth 677-86 1971 - McMichael AJ within 1974 Spirtas R Kupper LL An epidemiologic a cohort of rubber workers 1964-72 J | Oe study of mortality Occ Med 458-54 . Merliss RR treated rice and Japanese stomach cancer 3141-42 1971 Science 173 JACKSON 025775 335 employees Meurman LO Kilviluoto P Hakama M Mortality and morbidity of of anthophysllite asbestos mines in Finland Br J Ind Med 31 OO 105-12 Miller JA Carcinogenesis by chemicals 559-76 1970 an overview Cancer Research Ming SC Goldman H Gastric polyps a histologic classification and its relation to carcinoma Cancer 721-26 1965 Ming S Tumors of the Esophagus and Stomach Pathology Washington DC 1973 Armed Forces Institute of Monaco AP Roth SI Castleman B Welch CE Adenomatous Adenomatous polyps of stomach a clinical and pathological study of 153 cases 456-67 1962 the Cancer " Monk M Warshauser ME Stomach and colon Ricans in New York City and Puerto 1975 cancer mortality among Puerto Rico J Chron Dis 28 349-53 Morson BC Carcinoma arising from areas of intestinal metaplasia in the gastric mucosa Br J Cancer 377-85 1955.. - Intestinal metaplasia of the gastric mucosa 365-76 1955 Br J Cancer Munoz N Collelly R gastric cancer Time trends of intestinal and diffuse types of in the United States Int J Cancer 158-64 1971 Correa P Cuello C Histologic types of gastric carcinoma in high- and risk areas Int J Cancer 809-18 1968 Newell GR Waggoner DE Cancer mortality and environmental in the United States Lancet 763-68 1970 temper 2 tures Newill VA Distribution of cancer mortality among ethnic the white population of New York City 1953-1958 Inst 405-17 1961 subgroups of J Nat Cancer Pastor J as . Kato H Belsky J Serum pepsin and predictors of stomach cancer NEJM c : tubeless gastric 279-84 analysis Redmond CK Ciocco A steelworkers oven workers Lloyd JW Rush VI Mortality JOM621-29 HW term mortality from malignant neoplasms a 1972 study among of coke Smith EM Lloyd JW Rush HW term mortality study of steel- workers III Follow JOM513-21 1969 JACKSON 025776 336 Robbins SL Angell M 1971 Basic Pathology W. B. Saunders Co Philadelphia Robinson H term mortality study of steelworkers II Mortality by level of income in whites and whites JOM411-16 1969 Saccomanno G Archer VE Saunders RP Lung cancer of uranium miners on the Colorado Plateau Hlth Phys 1195-1201 1964 Schneiderman MA Digestive system cancer among persons subjected to occupational occupational inhalation of asbestos particles a literature review with emphasis on response Env Hlth Persp 307-311 1974 : Schonland M BradshaEw with reference to .82 1969 Upper alimentary the chewing tract cancer habit Brit in Natal J Cancer Indians 670- Schwartz D Flamant D Lellouch J Denoix P Results of a French survey on the role of tobacco particularly inhalation on different cancer sites J Nat Cancer Inst 1085-1108 1961 Seidman H Cancer death rates by site and sex for religious and socioeconomic groups in New York City Environ Res 234-50 1970 Selikoff IJ Churg J Hammond EC 22-26 1964 Asbestos exposure and neoplasia JAMA Hammond EC Churg J JAMA 104-112 1968 Asbestos exposure smoking and neoplasia , Churg J Hammond FC The occurrence of asbestos among insulation . workers in the United States Ann Y. Acad of Sc 139-55 1965 Selikoff IJ Epidemiology of gastrointestinal cancer 299-305 1974 Env Hith Persp Carcinogenicity Selikoff IJ Hammond EC Seidman H Arch Env Hlth 183- = 1972 of amosite asbestos Shamberger RJ Tytko S Willis CE Antioxidants in cereals and in food cancer preservatives and declining gastric cancer mortality Clev Clin Quart 199-24 199-24 1972 + - - nO: "ae Shettigara PT Morgan RW Asbestos smoking and laryngeal carcinoma Arch Env Hith 517-519, 1975 Sigurjonnson J Occupational variations in mortality from gastric cancer in relation to dietary differences Br J Cancer 651-56 1967 JACKSON 025777 337 Sivralla M Seppala K Atrophic gastritis as a possible gastric carcinoma and pernicious anemia results examinations Acta Scand 455-74 1960 precursor of of follow Smith RL Recorded and expected mortality among the Japanese of the United States and Hawaii with special reference to cancer J Nat Cancer Inst 459-73 1956 Staszewski J Haenszel W Cancer mortality among the Polish born in the United States J Nat Cancer Inst 291-97 1965 ,> McCall MG StenhousNeS Cancer mortality in 1962-66 among Polish migrants in Australia Brit J Cancer 599-610 1971 Stell PM McGill T 1973 Asbestos and laryngeal carcinoma Lancet 416-17 Stempien SJ Rodiles DH Cooper EJ Dagrad AE quantitative gastric acid measurements 1974 Clinical correlation with Am J Gastroent 261-69 Sterup K Mosbech J Sex ratios of gastric cancer related to site of the tumor Scand J Gastroen Suppl 87-89 1971 St^cks P Cancer mortality in cigarettes solid fuel 1970 relation tea and to national consumption of coffee Brit J Cancer 215-25 . Cancer of the stomach in the large towns of England and Wales 1921-39 Br J Cancer 147-57 1950 Davies incidence 14-24 RI Zinc of cancer 1964 and copper content of the stomach and of soils associated other organs Brit with the J Cancer Stocks P On the relations between atmospheric pollution in urban and rural localities and mortality from cancer bronchitis and pneumonia with particular reference to 4 benzopyrene berylium molydenum vanadium and arsenic Br J Cancer 397-418 1960 Stout AP Tumors of the Stomach Washington DC 1953 Armed Forces Institute of Pathology Terris M. Hall CE Decline in mortality from gastric cancer in native and foreign residents in New York City 155-62 1963 J Nat Cancer Inst Tomasulo J Gastric polyps histologic types and their relationship to gastric carcinoma Cancer 1346-55 1346-55 1971 JACKSON 025778 338 Tromp SW Diehl JC A statistical study between cancer of the stomach and 1955 of the soil possible relationship Brit J Cancer 349-57 Tulchinsky D Modan B Epidemiological aspects of canceorf the stomach in Israel Cancer 1311-17 1967 U. S. Cancer Mortality by County U. S. Dept of Health Education and Welfare Public Health Service National Institute of Health 1975 Vatn MH Gastric intrinsic factor secretion 38 1975 Scand J Gastroent 337- - Wagner JC Sleggs CA Marchand P Diffuse pleural asbestos exposure in the Western Cape Med 260-71 1960 mesothelioma and Province Br J Ind Wagoner JK Archer VE Carroll BE Holaday DA Lawrence PA Cancer mortality patterns among U. S. uranium miners and millers 1950 through 19 J. Nat Cancer Inst 787-801 1964 , Archer VE Lundin FE Radiation as the cause of lung cancer among uranium miners JEJM 181-88 1965 Miller RW Lundin FE Fraumen JF mortality among a group of underground 284-89 1963 Haij ME Unusual cancer metal miners NEJM Waterhouse JAH Cancer Handbook of Epidemiology and Prognosis Churchill Livingstone Edinburgh and London 1974 White C Eisenberg H ABO blood groups and cancers of the stomach Yale J Biol Med 58-61 1959 Winkelstein W Jr Kantor S ~ suspended particulate 1969 Stomach cancer air pollution - positive association with Arch Environ H1th 544-47 World Health Statistics Annual I Vital Statistics and Causes of Death World Health Organization Geneva 1976 Wynder EL Conference on the etiology of cancer of the gastrointestinal tract Cancer 1561-66 1966- Zacho A Cederquist C Nielsen J Larsen V Duration of smoking and quantity of tobacco usebdy patients with gastric cancer Acta Se Med Scand 45-48 1973 , Nielsen J Larsen V Smoking habits of patients with gastric cancer Acta Chir Scand 455-60 1971 JACKSON 025779