Document eZkqyBR7NKRL9aVB1d8myQem

FILE NAME: Goodyear (GY) DATE: 1977 Dec DOC#: GY049 DOCUMENT DESCRIPTION: Report - Lung Cancer Among Rubber Workers at the Goodyear Akron Plants: A Case-Control Study with Cover Memo Long, James rom: ent: To: Subject: Bogart, Russell Thursday, September 05, 2002 12:02 PM Long, James: Meadow, Richard: Steinwolf, Bonnie Goodyear In December 1977, University of North Carolina researchers prepared a report for the "The Joint URW-Goodyear Occupational Health Committee" titled "Lung Cancer Among Rubber Workers At the Goodyear Akron Plants: A CaseControl Study." This study discusses the then-existing knowledge of other risk factors for the development of lung cancer. With respect to the association between asbestos and lung cancer, the North Carolina researchers wrote, "a causal association between occupational exposure to asbestos fibers and lung cancer and pleural mesothelioma has been well-demonstrated." No evidence exists that Goodyear had ever disavowed In any manner this report prepared for, reviewed by, and presumably approved by and disseminated by the Goodyear Health Committee and Goodyear. Significantly, this Goodyear report cites primarily to Sellkoff and Churg's 1968 publication for this bolded proposition, and we can argue that Goodyear has acquiesced to the proposition that the link between asbestos and lung cancer and mesothelioma was established by 1968. Accordingly, we can argue that the jury can consider this adoptive admission as evidence showing that Goodyear should have warned users of Its gasket material at least by 1968..... i LUNG CANCER AMONG RUBBER WORKERS AT THE GOODYEAR AKRON PLANTS: A CASE-CONTROL STUDY Report Prepared for The J o in t URW-Goodyear O ccupational Health Committee by Elizabeth S. D e lz e ll, M.S.P.H. Occupational H ealth Stu dies Group U niversity of North Carolina Chapel H i l l , North C a ro lin a December, 1977 GY-20 GTR 000081 TABLE OF CONTENTS. Page SUMMARY............................................................................................... i L IS T OF TABLES ................................................. ............................... 1i i L IS T OF FIGURES ............................................................................... vi L IS T OF APPENDICES ...................................................... ................... v i i Chapter I. INTRODUCTION ............................................................................. 1 I I . LITERATURE REVIEW ................................................................. .. 3 A. D e s c r ip tiv e Epidem iologyand GeneralR isk F actors ........... 3 Extent o f the Problem ...................................................... 3 Trends Over Time in Lung CancerM o r t a lit y ...................... 4 G eographical V a r ia tio n .................................................... 5 Age ............................................................. 7 Sex .................................................................... 8 Race and E t h n i c i t y .......................................................... 8 Urban Environment .............................................................10 M ig r a tio n and N a t iv it y .....................................................12 Smoking ............................................................................. 15 Other Host S u s c e p t ib ili t y F acto rs .................................. 15 B. O ccupational F actors ........................................................ 16 General M ethodologie Problems in Occupational H ealth S tu d ie s ...............................................................16 Impact o f O ccupational Exposures on Lung Cancer M o r t a lit y ....................... .v .......................... 19 C. Lung Cancer F in d in g s in the RubberIn d u stry ....................34 D. H is t o lo g ic Types o f Lung Cancer .................................... 46 E. Summary o f the Background In fo rm a tio n and L it e r a t u r e Review ......................................................... 53 I I I . OBJECTIVES .................................. 55 IV . METHODS ......................................................................................58 A. R a tio n a le U n d e rlyin g the Choice o f StudyD esign .............. 59 B. Assum ptions R elevant to V a l i d i t y in Case-C ontrol Stu d ie s ..................................................... 62 GTR 000082 C. Design C o n sid e ra tio n s ................................................. 64 D escription o f the Sourceand Sample Populations ...6 4 V a lid ity with Respect to Se le ctio n o f Cases and C o n tro ls .............................................................67 V a lid ity with Respect to Control o f P o te n tia lly Confounding V a r ia b le s .............................................69 D. Sources and Adequacy o fData ....................................... 73 E. Analyses ...................................................................... 78 V. RESULTS AND DISCUSSION ........................................................88 A. R e su lts o f the M atching Procedure ............................ 88 B. C o m p arab ility o f Study Su b je cts w ith Respect to Control V a r ia b le s ...............................................88 C. Place o f B ir t h ............................................................ 91 D. R e su lts o f the O T G -S p e c ific A n aly se s '....................... 93 E. Summary o f the R e s u lt s o f the O T G -S p e c ific Analyses. 102 V I. CONCLUSIONS .......................................................................... 131 REFERENCES ............................................................................136 A PPEN D ICES............................................................................ 143 GTR 000083 Elizabeth Sills Delzell. Lung Cancer Among Rubber Workers: A Case-Control Study. Under the direction of Dragana Andjelkovich and Herman Tyroler. The following report is an adaptation of the thesis submitted to the faculty of the University of North Carolina in partial fulfillment of the requirements for the degree of Master of Science in Public Health in the Department of Epidemiology. SUMMARY Previous assessments of the effects of work--related exposures in the rubber industry on lung cancer mortality have been inconclusive but suggest that an increase in lung cancer risk may be associated with the history of employment in certain work areas within the rubber industry. Specifically, workers with exposures to processes involving the compounding and mixing of raw materials and the curing of green rubber products have been shown by more than one Independent study to sustain a greater than expected lung cancer mortality. These findings have not been consistent across all studies of mortality among rubber industry employees. The main objective of the present study was to identify groups of rubber industry workers, categorized according to their work experience, who may be at increased risk of dying of lung cancer. In addition to evaluating the lung cancer risk associated with employment in the compounding and mixing and curing work areas, the present study sought to ascertain the extent of associations, if any, between lung cancer mortality and history of employment in other work areas. - A matched case-control study design was used .to identify possible high risk work areas at an Akron, Ohio rubber plant. The detailed employment records of 121 white male lung cancer cases and 448 controls, matched on race, sex, year of birth, and year of first hire, were examined. Effect measures were estimated for 19 work areas. Mean durations of employment and mean ..years of first employment in each work area were also compared for cases and controls. -iGTR 000084 Results indicate that, for the population of rubber workers studied, there is no association between risk for lung cancer and history of employment in the Compounding and Mixing and Curing work areas. A statistically significant elevation in risk was detected for the Reclaim Operation work area, for those employed for 5 or more years. Furthermore, among cases and controls employed in Reclaim Operation for at least 1 month, cases worked for twice as long, as controls, on the average, in this work area. Study findings also suggest that there may be weak associations between a history of employment in work areas designated as Chemicals, Pliofilm, and Special Products Manufacture. Further epidemiologic and environmental research is needed to identify specific respiratory system hazards, if any, in these work areas. ' Information regarding cigarette consumption and several other known risk factors for lung cancer was not available for the subjects of the present study. Because it was not possible to control for the effects of these potential confounders, the results of this study should not be regarded as conclusive. t 3 LIST OK TABLES Table Page 1 . Latent P er io ds of O c c u p a t i o n a l Lung C a n c e r s ...... 19 2-a. O cc u p a t i o n a l Lung Cancer: I n d u s t r i a l E x p o s u r e s to In organic C o m p o u n d s ................................ 20 r 1 CM Occupational Lung Cancer: Industrial Exposures to Organic C o m p o u n d s ...................................30 3. Lung Canc er M o r t a l i t y in the U.S. and B r it i s h Rubber Industries: O v er a l l F i n d i n g s . . . . ; . . . . ..... 36 4. Respiratory Cancer Mortality in the U.S. $ Bri tish Rub b e r Industries: R e l a t i o n s h i p to Specific J o b s ........................................... 39 5. H i s t o l o g i c Types of Lung Cance r: E s t i m a t e s of Rel at iv e F re qu e n c i e s of Four M a j o r T y p e s .......... 48 6. Chemicals Present in the Compounding and Mixing and the Curing O T G 's that are Known R es pi r a t o r y Sy st em H a z a r d s ........................... 76 7. Results of the Matching Procedure for the Rubber W or k e r Lung Can ce r C a s e - C o n t r o l S t u d y ............ jgg 8. Selection of Rubber Worker Lung Cancer CaseControl S tud y Sugjects: R e a s o n s for E x c l u s i o n s .-108 9. Re su l ts of the M a t c h i n g P r o c e d u r e f o r the R u b b e r W o r k e r L u n g Cancer C a s e - C o n t r o l S t u d y : ..... IQ.9 10. Selected Characteristics of Lung Cancer Cases and C o n t r o l s ........................................... H O 11. Frequency Distribution of Year of First Hire for Lu n g Can cer Cases and C o n t r o l s ..................... Ill 12. Erequency Distribution of Duration of Rubber Industry Employment for Lung Cancer Cases and C o n t r o l s ................................................ 112 13. Place of Birth D i s t r i b u t i o n for R u bber W o r k e r Lung Can cer Cases and C o n t r o l s ..................... 113 14. P l a c e of B i r t h D i s t r i b u t i o n for N a t i v e - B o r n Rubb er W o r k e r Lung C a n c e r Cases and C o n tr o l s . . . .113 ' --iii-- A; GTR 000086 LIST OF TABLES, cont. Table Pag 15. Place of Birth: C h i - S q u a r e Test of Association, Southern St at e s vs. All Other P l a c e s of Birth. JL14 16. Place of Bir th: C h i - S q u a r e Test of As so ci a ti o n , N o r t h eas te rn State s vs. All Other Place s of B i r t h ................................................. .114 17. Number and P e r c e n t of Lung Cancer Cases and Controls Ever Employed in each of 20 OTG's. 116 18. Number and Per Cent of Lung Cancer Cases and Controls Employed for at Least 2 Years in each of 19 O T G 's .......................................... 19. Number and Per Cent of Lung Cancer Cases and Controls Employed for at Least 5 Years in each of 19 O T G ' s ................ , ......................... 118 20. Odds Ratios for E m p l o y m e n t in 19 O T G' s bv Three M i n i m u m D u r a t i o n of Em p l o y m e n t C a t e g o r i e s ............................................ 119 21. Odds Ratios for T h o s e Empl oy ed, C o m p a r e d to Those' Never Em p l o y e d , for 19 OTG's by Three M i n i m u m D u r a t i o n of E m p l o y m e n t C a t e g o r i e s ......120 22. Odds Ratios for M a t c h e d Gro up s for 19 OTG's by Three Minimum Duration of Employment C a t e g o r i e s .............................................121 23. Odds Ratios for M a t c h e d Group s for 19 OTG's by Two M i n i m u m D u r a t i o n of E m p l o y m e n t C a t e g o r i e s . .122 24. Odds Ratios for 19 OTG's: Exposed Defined' as Employed in the OTG for at Least 1 Month, 2 years, or 5 years and First Employed in the OTG 15 or Mo re Y e a r s P r i o r to Death; N o t E x po s e d Defined as Employed in the OTG for Less Than 1 .. Month or F i r s t E m p l o y e d in the OTG F e we r Th a n 5 Years P r i or to D e a t h ................................ 123 25. Mean Ye ar of First E m p l o y m e n t in 19 OTG's for Cases and Controls Employed for at Least One Month in the O T G of I n t e r e s t ...................... 124 26. Mean Year of F i r s t E m p l o y m e n t in 19 OTG's for Cases and Controls Employed for at Least 2 Years In the O T G of I n t e r e s t ........... .......... 125 27. Mean Year of F i r s t E m p l o y m e n t in 19 OTG's for Cases and Controls Employed for at Least 5 Years in the OTG of I n t e r e s t ...................... 126 -i vGTR 000087 LIST OF TABLES, cont. Tabl Page 28. Mean Duraticrn of E m p l o y m e n t in 19 OTC's for Lung Cancer Cases and Controls, Based on the Experi e n c e of Workers H aving at Least 1 Month of E m p l o y m e n t in the O T C ' s ..............197 29. Summary of M a n t e l - H a e n s z e l Odds Rat io s for C o m p o u n d i n g and M i xi n g and C u r i n g ............ ^28 30. Summary of M a n t e l - H a e n s z e l Odds Ratios for Re c l ai m O p e r a t i o n (OTG 1 4 ) ..................... 129 31. Summary of M a n t e l - H a e n s z e l Odds Ratios for Chemicals (OTG 15), Pliofilm (OTG 16), and Special P r o d u c t s M a n u f a c t u r e (OTG 1 9 ) . . ..... 130 -V- nnnnn o LIST OF FIGURES Figure Page 1. M o r t al i ty ex per i e n c e of the 1964 Cohort of rubber w o r k e r s ..................................... 65 2. Da ta layout for c a l c u l a t i o n of the c r u d e odds ratio (OR) in c a s e - c o n t r o l s t u d i e s ......... 81 3. L a yo u t and n o t a t i o n for data fr o m c as econtrol studies w i t h .4:1 m a t c h i n g and di ch ot o mo us o ut co me ande x p o s u r e .................. 83 -vi GTR 000089 LIST OF APPENDICES A - l . Occupational T i t l e L i s t ...................................... A-2. Occupational T i t l e Group D ic t io n a r y .................. A-3. Summary D e s c rip tio n o f O ccupational T i t l e Groups B. A d d itio n a l Formulae Used in A n a ly se s ................. --v i i -- GTR 000090 the rubber industry is indicated, in order to clarify the interpretation of these results. Thus, it is the goal of the present investigation to identify sub-groups of workers who may be at high risk of lung cancer mortality by virtue of their employment experiences in the rubber industry. In order to obtain a valid evaluation of the possible association between rubber industry employment and lung cancer mortality, other known disease determinants that may also be related to job patterns within the rubber industry must be considered. These known correlates of or risk factors for lung cancer will be reviewed in Chapter II. GTR 000092 CHAPTER II LITERATURE REVIEW A. Descriptive Epidemiology and General Risk Factors 1. Extent of the Problem Lung cancer^, as a cause of cancer death in the United States, ranked first among males and fourth among females in 1969 and accounted for approximately 75,000 deaths in 1974 (11,12). For the latter year, estimates based on the Third National Cancer Survey indicated that there were 83,000 new cases of lung cancer in the United States (11). Lung cancer is a rapidly fatal disease, with a 5-year survival for all stages of 8% for males and 12% for females (1). Thus, mortality data for lung cancer yield conservative but not greatly distorted estimates of the incidence and impact of lung cancer. In 1967, the following age-adjusted mortality rates per 100,000 for cancers of the lung, bronchus, and trachea 2 were noted (13) : 19.7 for white males; 22.1 for nonwhite males; 2.9 for both white and nonwhite females. The rates ar6 applicable to cancers of the lung, bronchus, and trachea that were designated as primary. Mortal ity rates that combine lung cancers specified as primary with those The terms "lung cancer" and "cancer of the lung" used in this paper refer to all cancers of the lung, bronchus, and trachea, International Classification of Disease (ICD) (Eighth Revision) #162, unless otherwise stated. 2 Directly adjusted, using the 1960 U.S. population as a standard. 6TR 000093 4 unspecified as primary or secondary are considerably higher: for example, 49 per 100,000 for white males and 54 per 100,000 for nonwhite males. Murray and Axtell (14), using U.S. Vital Statistics data for 1968, calculated work-years lost due to lung cancer deaths based on life expectancy coverage for 'the working years included in the age interval of 20-65 years. It was estimated that lung cancer deaths in 1968 accounted for 222,948 lost work-years for white males and thus had the greatest impact on the male labor force of all cancer causes of death. It is apparent that lung cancer has a substantial impact on both the general and working populations in this country in terms of overall mortality and work years lost. 2. Trends Over Time in Lung Cancer Mortality Lung cancer mortality rates rose rapidly for all race-sex groups between 1950 and 1967 in the U.S. (13). Figures for other countries indicate that this increase over time has been consistent in all parts of the world studied (15). For the U.S., the age-adjusted lung cancer mortality rates for white males and females and for nonwhite females more than doubled between 1950 and 1967, and a three-fold increase was sus tained by black males during the same time period (16). Higgins (17) examined 1940 through 1967 time trends in age-specific respiratory cancer mortality rates and reported increases for ages 35-74 for all race and sex groups. Among U.S. white males, ages 35-44, lung cancer mortality rates increased at a constant rate between 1940 and 1967, while in the three older 10-year age groups, lung cancer mortality rates increased at a declining rate. The age differential in the fate of increase noted 'by Higgins may be indicative of either additional or higher dose exposure for the 5 younger age groups or of relatively earlier exposure to carcinogenic or promoting agents. Higgins also reported that, in England and Wales, lung cancer mortality rate per 100,000 increased from 39.08 in 1950 to 67.72 in 1964 for males and from 5.8 to 9.7 for females. The comparable figures for the U.S. white population are, for males, 18.44 in 1950 and 36.86 in 1964 and, for females, 3.83 in 1950 and 5.83 in 1964. These changes correspond to the following percent increases in lung cancer mortality rates: in England and Wales, 73.3% for males and 67.2% for females; in the U.S., 99.9% for males and 52.2% for females. It should be noted that the difference between England and Wales and the U.S. in the percent change in male rates--73.3% compared to 99.9%--may be par tially accounted for by the fact that the U.S. rate in 1950 was lower than the rate for England and Wales in 1950, and, consequently, a smaller absolute increase in the U.S. lung cancer rate over the 14-year time period produced a larger percent change. However, for both countries, the observed temporal trends probably reflect changes in the prevalence of environmental agents that have an impact on lung cancer development and, in particular, change in tobacco consumption habits, occupational exposure patterns, and urban air pollution levels.' - 3. Geographical Variation According to estimates for 1966-67, lung cancer mortality rates show extensive geographical variation (15). In general, rates are highest in the highly industrialized countries of northern Western Europe and relatively low in Southern Europe (Italy and Portugal), Japan, and Chile. However, lung cancer rates may exhibit considerable variation within such geographical regions. For example, while Finnish males had a high GTR 000095 6 age-adjusted lung cancer death rate in 1966 through 1967 (61 per 100,000), Norwegian males had the third lowest rate among 25 countries (14.9 per 100,000). In a rank ordering of lung cancer mortality rates among males for 25 countries (15), England and Wales were ranked highest, the U.S. ranked ninth and tenth for nonwhites and whites respectively, and Portugal was ranked lowest. A greater than seven-fold variation in rates between the lowest and highest ranked countries was noted. Doll (18), using approxi mate cumulative incidence rates, estimated a world-wide range in risk of developing lung cancer for populations under 75 years of age. He reported that this risk was 11% in England, while for Nigeria, it was 0.1%-- a 100 fold variation. This estimate must be viewed with caution, since the cumulative incidence (risk) for Nigeria may be underestimated because of inaccurate or incomplete disease reporting practices. Some of the observed variability in lung cancer mortality rates between the sexes within a given country and among various countries within larger geographical groupings may certainly be attributable to differences in completeness of lung cancer detection and diagnosis and, possibly, in enumeration of the population-at-risk forming the denominator of the rates. However, geographical variation may also reflect differences in exposure to lung carcinogens, in competing causes of mortality, or in host susceptibility. Geographical variation in lung cancer mortality patterns is also apparent within the U.S. (19). For white males, areas having age-adjusted rates that are significantly high compared to the U.S. as a whole include urban regions in the Northeast; parts of Texas, Louisiana, Mississippi, Alabama, and Florida that border the Gulf of Mexico; and parts of eastern GTR 000096 Florida, Georgia, and South Carolina that lie along the Atlantic coast. This pattern is similar, although less pronounced, for white females. The geographical variation in lung cancer mortality in the U.S. may be partially attributed to effects of urban air pollution and to differences in employment in heavy industry and in consumption of cigarettes. The explanation for the high lung cancer death rates experienced in the Gulf of Mexico and southern Atlantic coast regions is obscure. However, a recent study by Blot and Frauraeni (20) suggests that the presence in these areas of petroleum, ship building, and paper industries is corre lated with higher than expected lung cancer death rates for both white and nonwhite males. Employment in these industries may involve exposure to lung cancer hazards. 4. Age Cross-sectional analyses of age-specific lung cancer death rates indicate that rates for males increase for each five-year age interval after the age of 35, up to age 70, then decline rapidly (21-23). The age * curve for females does not exhibit this pattern: rates for females increase steadily with age, up to the oldest ages, and the rate of increase is much less for females than for males. When analyses of male lung cancer death rates by birth cohort are performed, it is apparent that both male and female rates increase with each increment in age. Furthermore, each successive birth cohort experiences a higher lung cancer death rate. Thus, at a given age-at-death, the rates increase for each birth-year cohort. This cohort effect has been ascribed to intensified smoking among successive birth cohorts. GTR 000097 Lung cancer incidence and mortality rates for males are consistently higher than female rates in the U.S. and in other countries studied (13,15,24). Cancer morbidity surveys in the United States indicated that male lung cancer rates for both whites and nonwhites were approxi mately three times higher than rates among females in 1937, 4.5 times higher in 1947, and about 5 times higher in 1969. Comparisons of age- adjusted lung cancer mortality rates for the years 1950-1967 reveal that sex ratios increased for whites and nonwhites through 1960. Since 1960, sex ratios have been decreasing steadily. This decrease is due to a greater percent change in lung cancer death rates for females than for males during this time period. . The lower lung cancer rates observed among females may be attribut able to several factors. Among these are male/female differences in occupational exposure to respiratory system hazards and in cigarette smoking habits; it has been suggested that females smoke fewer cigarettes, inhale less intensely, and begin smoking at later ages than males (25). 6 . Race and Ethnicity . ' _ Prior to 1960, lung cancer mortality rates in the U.S. for nonwhite \ males and females were below those for whites. In .1960 and in subsequent years, nonwhite rates have been higher than rates among whites (13). In 1969, lung cancer incidence was 84.7 per 100,000 for nonwhite males, compared to 68.9 per 100,000 for white males. The corresponding rates per 100,000 for females were 18.2 for nonwhites and 13.5 for whites. It is apparent that the race differential is greater for males than for females, possibly reflecting greater differences in exposure to carcino gens and/or in susceptibility among nonwhite compared to white males GTR 000098 9 than among nonwhite compared to white females. Several other ethnic groups in the U.S. have lung cancer mortality experiences that differ from that of the general U.S. population. Fraumeni and Mason (26) evaluated lung cancer mortality among Chinese Americans. Their analysis was based on death certificates obtained for the years 1950-1969 and Census information for 1950, 1960, and 1970. They determined that age-adjusted lung cancer death rates were signifi cantly higher for both male and female Chinese Americans than for both the general white and nonwhite populations of the U.S. The excess was greatest for Chinese American females. The magnitude of this excess- twofold for females-- is not large and may have resulted partly from underenumeration by the Census of the Chinese American population. Fraumeni and Mason were unable to differentiate between immigrant and U.S.-bom Chinese. Also, no information on cigarette smoking habits of the Chinese Americans was available. Therefore, it is difficult to interpret the results of this study. Creagan and Fraumeni (27) studied the 1950-1967 cancer mortality among American Indians, again using death certificates and Census infor mation. Comparisons of observed numbers of lung cancer deaths with expected numbers based on the lung cancer mortality experience of the general U.S. population indicated a substantial deficit of lung cancer among both male and female Indians. The authors suggested that differ ences in cigarette smoking between Indians and the general U.S. popu lation might have accounted for the observed deficit. Finally, Buell, et al. (28) identified an excess in lung cancer mortality among female Mexican immigrants in the U.S. The observed excess was largely confined to those who had spent their early life in GTR 000099 10 Mexico and was ascribed by the investigators to heavy cigarette smoking by these women. 7. Urban Environment Numerous studies carried out in Great Britain and the U.S. indicate that lung cancer mortality rates are higher in urban than in rural areas (29-34). The lung cancer excess observed in urban areas has been attributed partially to the higher degree of air pollution found in the urban environment. Interpretation of the findings of these studies has been hindered by lack of or poor quality of information pertaining to tobacco consumption, changes in residence, and environmental measurements. However, several studies of lung cancer mortality have examined both place of residence and smoking habits of study subjects (30-32). Results consistently indicate that at each level of smoking, age-adjusted lung cancer death rates are higher for residents of urban areas and that the urban effect is greatest among nonsmokers. For example, in a prospective study of 187,783 U.S. white males followed between 1952 and 1955, Hammond and Horn reported the following results: a) age-adjusted lung cancer mortality rates per 100,000 among nonsmokers increased from 4.7 for residents of town's having a population of fewer than 10,000 to 14.7 for residents of cities having a population in excess of 50,000; and b) for smokers, rates increased from 71.7 to 85.4 per 100,000 over the same population gradient. In 1958, Haenszel and his coworkers (30) were able to collect data on both smoking habits and duration of residence for 2,191 white male lung cancer deaths occur ring during that year and for a representative national population sample of approximately 30,000 white males, ages 35 and over. They found GTR 000100 11 that the lung cancer excess for residents of urban versus rural areas increased with duration of residence. Furthermore, among lifetime residents of urban or rural areas, the excess in lung cancer mortality for urban residents was present for every smoking category and was greatest for non-smokers and smallest for smokers of more than one pack per day. Neither of the above-mentioned studies was able to take occu pation of study subjects into account. Hammond (35) reported the 'results of a study of 1,000,000 men and women in 25 states followed prospectively for six years, starting in 1959. Information on smoking habits, place of residence, and occupation was collected. In order to evaluate the effects of urban residence, Hammond confined analyses to male subjects who had lived for at least ten years in their place of residence at the beginning of the study. Age and smoking-adjusted lung cancer mortality rates were calculated for men occupationally exposed and not occupationally exposed to dusts, fumes, vapors, gases, or X-rays, separately for urban and non-urban residents. Results indicated that, when occupational exposure to respiratory hazards ^ was taken into consideration, only slight residual differences in lung cancer mortality attributable to urban versus rural residence were detectable. This study emphasizes the importance of controlling not only for smoking and age, but also for occupation, when examining the - association between urban residence and lung cancer mortality. ' The effects of urban air pollution on lung cancer occurrence remain unclear. If the association is real, its strength is clearly much smaller than that of the relationship between lung cancer and smoking. Several studies (33,34) have indicated that the effect of urban residence is considerably less in females than in males. Although these studies did GTR 000101 12 v not control for smoking, this observation provides additional evidence for Hammond's suggestion that urban/rural variation in lung cancer mortality may be due to occupational factors. 8. Migration and Nativity Studies of the cancer experience among migrants are of interest, since they potentially provide a means of determining whether environ mental or genetic factors predominate in the etiology of a given cancer. Mortality rates for cancers whose development is influenced by environ mental factors will assume a level for migrants that is intermediate between the death rate in the country of origin and that in the country of destination, reflecting the influence of environmental factors present both in early life, prior to migration, and in subsequent time periods (36) . In general, lung cancer death rates for migrants are intermediate between the rates in their new and former countries (37) . This has been demonstrated for British migrants to South Africa (38) . Lung cancer mortality rates of native-born residents of South Africa, British-born residents of Great Britain, and British immigrants to South Africa were compared. The rate for immigrants was intermediate between the high British rate and the* lower South African rate. No evidence`for a rela- tively decreased amount of smoking among British migrants could be found to account for their lowered (in comparison to British rate) lung cancer rate. Reid et al, (39) compared the rates of Norwegian and British immi grants to the U.S. with rates for the native-born U.S., Norwegian, and British populations. 1960-62 rates for Norway and Britain and 1959-61 ^^rates for the U.S. were used for these comparisons. For the British immigrants, age-adjusted lung cancer death rates were higher than U.S. GTR 000102 rates and lower than British rates. Lung cancer mortality rates in Norway were lower than in the U.S. Norwegian immigrants were shown to have lung cancer mortality rates higher than the Norwegian rates but lower than the U.S. rates. Haenszel (37), using all lung cancer deaths occurring in 35 states in 1950, found intermediate rates for male migrants from Italy, Germany, Sweden, Norway, and England and Wales. No consistent pattern was ob served for females, and no cigarette consumption data were available. Mancuso (40) compared the 1947-51 lung cancer mortality experience of male foreign migrants to Ohio with that of white males in Ohio. Again, intermediate rates for German, British, and Italian male immigrants were reported. The effects of migration within the U.S. on lung cancer mortality have also been evaluated. Haenszel (30) demonstrated an excess in lung cancer mortality for white males who had moved from rural to urban areas in the U.S. Mancuso (41) analyzed the mortality experience of migrants * ^ to Ohio. The study included all deaths reported for Ohio residents dying in the U.S. and Canada between 1959 and 1968. Directly age- adjusted rates were computed, using the 1960 general U.S. population as the standard. Lung cancer death rates were.found to be higher than rates for native b o m Ohio residents in the following groups: a) white male and female residents of Ohio who were b o m in the South or the Northeast; b) non-white male and female residents of Ohio who were b o m in the South or Northeast. The differences between lung cancer death rates for migrants and those for native-born Ohioans were greater for males than for females and were particularly striking for non-white males. 14 The observed differences in lung cancer mortality between migrants from the Northeast and native Ohioans suggest that environmental factors present in early life influenced the subsequent mortality experience of migrants from the Northeast, since this region has a high lung cancer death rate. However, lung cancer mortality rates in much of the South are lower than in the ILS. and are not significantly different from rates in most of Ohio. Mancuso suggests that the unfavorable lung cancer mortality experience of migrants to Ohio from the South may be attributable to the differential employment of migrants in hazardous occupations and heavy industrial settings. However,, other explanations for the apparent elevated lung cancer mortality among the southern migrants cannot be ruled out. The possibility exists that individuals ^ |igraing. from the South to Ohio were more heavily exposed to environ mental carcinogens in their native southern states than either the population remaining in the South or the native Ohioans. In addition, if one assumes that economic deprivation provides the impetus for migra tion, it is possible that migrants, as a group, compared to native Ohioans, may be of lower social class, may be in poorer general health because of nutritional dficiences and lack of medical care experienced in their birth states, and may have less access to health care facili ties and fewer social supports in theiT adopted state than do native Ohioans who do not emigrate. These factors may differentially influence the susceptibility of southern migrants to Ohio. Mancuso and other investigators have not considered the possible impact of these variables on the lung cancer mortality experience of migrants. 1 GTR 000104 16 several studies have shown that lung cancer patients have higher levels of AHH inducibility than do normal controls (45-47). The possibility that AHH inducibility, as well as other genetic biochemical markers, may prove to be useful in identifying individuals who are at particu larly high risk for lung cancer has been discussed recently by Mulvihill (48). B. Occupational Factors It is recognized that exposures to environmental carcinogens, promoters, and respiratory system irritants encountered in various occu pational settings can substantially increase the risk for lung cancer among industrial workers, especially among those who smoke (3). A number of occupational lung cancer studies have provided assessments of the association between lung cancer mortality and work-related exposures. These studies, and occupational studies in general, have been hindered to varying degrees by lack of adequate data, as discussed below. First, there are frequently insufficient numbers of employees-atrisk and employee deaths to yield stable risk estimates. Second, infor mation pertaining to the smoking and demographic characteristics of the study population may not be obtained. Some of these factors may be major disease determinants and may, therefore, distort attempts to evaluate associations between occupational exposures and health conse quences. The validity of findings .may be seriously questioned when measurement and control of the effects of these potential confounders are neglected. Third, ascertainment of the population-at-risk, both exposed and unexposed, and of subsequent mortality may be incomplete. ^x d nnn m e 17 Again, the validity of estimated effect measures derived from cohort studies may be challenged on the basis of this limitation. Besides the problems of incomplete identification of workers-atrisk and attrition caused by unsuccessful attempts to trace the mortality experience of all study subjects, several additional factors that potentially lead to dilution of the estimated effects of ex posure derived from studies of various occupational groups and that may contribute to great variation in-effect measures over the relevant studies are often not considered. _ First, detailed employment information, permitting accurate assessment of exposure, is often unavailable. Lack of exposure information, in terms of qualitative and quantitative environmental measurements, as well as changes in exposures over time, may lead to misclassification of workers by exposure. This inability to distinguish between lightly and heavily exposed workers and the resulting misclassification of workers according to the exposure of interest may bias measures of effect towards the null. Furthermore, failure to take into account the possible effects of other or multiple exposures often present in complex industrial environments may restrict- inferences that can be made regarding the causal' nature of any observed association between the disease and the exposure of primary interest. Second, adequate follow-up or lapsed amount of time between first exposure and initiation of the study to allow lung cancer or other cancers to occur may not be provided. As a consequence of this problem, the impact of the exposure on the health effect of interest is likely to be underestimated. Provision of a sufficiently long follow-up period is partly dependent on accurate knowledge regarding the length of the latent GTR 000107 18 period or induction time of occupational lung cancers. Estimates of the latent period for lung cancer, based on the findings of epidemiologic studies, are extremely variable, as demonstrated in Table 1. Values in this table derive from the results of more than one study of each exposure mentioned. The fluctuations in estimated mean latent period and the width of the reported ranges may reflect both differences in methodology of detecting cases and determining exposures and varia tion in potency and dose of carcinogens constituting the "exposure" in each study situation. Lack of accurate knowledge of the induction times of lung cancers associated with work exposures prevents the exclusion of both irrelevant exposures and irrelevant cases and the identification of the time period during which exposures of etiologic significance could have taken place. An accurate calculation of effect measures, such as relative risk for exposed versus non-exposed subjects, is possible only when appropriate length of follow-up, taking latent period into account, has been provided and when irrelevant cases or exposures have been excluded. Finally, when the occupational group of interest cannot be separated into exposed and unexposed subgroups, an external population must be sought in order to make comparisons and to derive estimates of effect of the exposure of interest. The magnitude of estimated effects may vary according to the particular population used for comparisons. Since working populations are usually healthier than a general popula tion, choice of the general population of a country or a state as a comparison group may contribute to a dilution of the observed effects of exposure. GTR 000108 19 TABLE 1 LATENT PERIODS OF OCCUPATIONAL LUNG CANCERS . Agent Average Latent Period (years) Range of Latent Period (years) Chromates 15 5-47 Nickel 22 6-30 Ionizing radiation 25-35 7-50 Asbestos 18 15-21 Tar Fumes 16 9-23 BCME (bis-chlororaethyl ether) 15 10-24 SOURCES: Kotin, P. (Ref. #36) Weiss, W. and Figueroa, W.G. (Ref. #73) Because of the above-mentioned and other limitations, the findings of occupational lung cancer mortality studies should be interpreted with caution. The results of a number of such studies will be reviewed below. First, findings regarding lung cancer mortality in industries involving exposure to inorganic substances will be discussed. These industries include the following: chromate production and processing, nickel refining, cadmium smeltering, copper smeltering, lead sraeltering, iron ore mining, insulation and other work involving asbestos. Also included in this section will be uranium mining. Major findings rele vant to the lung cancer mortality experience of workers in these occu pations are presented in Table 2-a. GTR 000109 ) ) ) TABLE 2-a OCCUPATIONAL LUNG CANCER: INDUSTRIAL EXPOSURES TO INORGANIC COMPOUNDS Industry/Occupation Suspected Agents Experimental Data* Estimated Magnitude of Excess Reference # Chromate Production 8 Processing (Pigments Industries) lead chromate .......... + zinc c h r o m a t e .......... + SMR = 486-186S SMR = 3800 SMR = 350 49,50 *52 51 Nickel Refining Nickel Carbonyl ........ + SMR = 130-1050 54 other Ni compounds . . . + SMR = 600-800 55 Cadmium Smelter Workers Cadmium oxide dusts . . . + Cadmium sulfate ......... + SMR = 235-299 (for 56 workers having 30+ years since first hire) Copper Smeltering; Arsenic compounds . . . . - Insecticide 8 Pesticide (As-Ow, Na arsenite, Producers 8 Users Pb arsenite) SMR = 450-800 (for 58 workers exposed 15+ years) SMR = 350 S9 Lead Smelter 8 Battery Plant Operators Various: Pb, As 8 Cd compounds; S0_ . . . . + * (for Cd) SMR = 148 (Smelters) 61 SMR = 132 (battery plant) 61 O I/ -*.!*> . Jl. . .. TABLE 2-a, continued Industry/Occupat ion Iron Ore (Hematite) Miners Suspected Agents Experimental Data* Estimated Magnitude of Excess Iron Oxides ........ Radiation .......... . + PMR = 174 Reference 63 Asbestos Workers Talc Miners Millers Uranium Miners Asbestos fibers . . . . . + Talc dust ` (asbestos fibers) . . . + Radon daughters ,214n. 214n 218n . ( Pb, Po, Po) . + SMR = 760 SMR = 417 PMR = 320 SMR = 900 64 65 66 68,69 * A "+" indicates that results of experimental tests are positive for the agent of interest* a indicates negative results. ' 22 Taylor and his co-workers (4 9 >50) studied the mortality experience of 1,213 male chromate workers at three plants. The study population consisted of workers who were employed between 1937 and 1940. Not all workers employed prior.to 1937 were included in this investigation. The follow-up period for subsequent mortality among the workers was approximately twenty years, through 1960. There were 71 respiratory cancer deaths. Standardardized Mortality Ratios (SMR's), with expected numbers of deaths calculated on the basis of U.S. male rates, were 468 for workers employed between 1937 and 1940 and 1864.5 for those first employed between 1941 and 1945, indicating an`approximately 4.5- to 18.5-fold excess for these workers. . Bidstrup and Case surveyed 724 British chromate workers employed at three plants in 1949 (51). Seven hundred twenty-three were found to be free of lung cancer and were followed for approximately 5.5 years, through August, 1955. There were 12 lung cancer deaths, compared to 3.3 expected, based on the mortality experience of the male population of England and Wales. Thus, the SMR was 350. Furthermore, for seven (58%) of the deaths, the age at death was less than 55. It is unusual to find this large a proportion of young decedents among a group of lung cancer deaths. * Langard and Norseth (52) studied 24 workers who had been employed for at least three years in a Norwegian chromate pigment producing plant and noted an excess in lung cancer mortality. However, their observation was based on only three lung cancer cases. Lead, zinc, and calcium chromate have been found to be carcino genic experimentally (53). In addition, these compounds are pulmonary irritants. The lung cancer risk for users of chromate-containing pig ments has not yet been evaluated. r.TR 000112 23 Doll (54) followed the mortality experience of 845 nickel refineryworkers, having at least five years of employment and first employed prior to 1945, from 1939 through 1966. Annual age-and cause^specific rates for the population of England and Wales were applied to the manyears at risk of the workers to derive SMR's. SMR's for those first employed before 1925 were approximately 950, while for workers first employed after 1925, the SMR was 130. Nickel carbonyl, which has been found to be a pulmonary carcinogen in rats, was manufactured at the plant both before and after 1925. The authors concluded that nickel carjbonyl was not solely responsible for the excess in lung cancer deaths among the refinery workers. Other nickel, compounds, alone or in combination with arsenic present as an impurity in the refining process prior to 1925, may have contributed to the observed elevated lung can cer mortality. Pederson (55) investigated 1,916 Norwegian nickel refinery workers who had first been employed before 1961 and who had worked in the refinery for at least three years. This population was followed from 1953 through 1971. Expected numbers of deaths were calculated on the basis of national age- and calendar-year-specific rates. SMR's for various jobs were 600-800. The greatest excess'(SMR=1000) was experienced by workers first employed before 1945. In 1950, production changes that may have influenced exposures within the refinery took place. There is some evidence that cadmium smelter workers may also be at increased risk for lung cancer, compared to the general population. Lemen et al.(56) followed 292 white male cadmium smelter workers, having a minimum of two years of employment and first employed between GTR 000113 24 1940 and 1969, through 1973. Their SMR for lung cancer was 235. This SMR is based on only two observed deaths and cannot be regarded as strong support for the hypothesis that cadmium exposure increases the risk for lung cancer. Arsenic, zinc, lead, and copper were also pre sent as impurities in the cadmium smelter operations. Copper smelter workers have been found to have excessive lung cancer. It is thought that the observed lung cancer excess may be caused by exposure to arsenic, and nine out of eleven epidemiologic studies of lung cancer mortality and exposure to arsenicals via smeltering and formulation and use of pesticides and insecticides contain ing arsenic have provided support for this hypothesis (57) . Lee and Fraumeni (58) compared the 1938-63 mortality experience of 8,047 white male smelter workers exposed to arsenic trioxide with that of a referent group consisting of the white male population residing in the same geographical region. Significant excesses in lung cancer mortality among smelter workers were detected for each duration of employment category. Rates more than 4.5 times higher than expected were observed for workers employed for at least 15 years, with the 15th year completed before 1938. Work areas within the smelter were grouped according to intensity of exposure to arsenic into heavy, medium, and light exposure areas. An SMR of 800 was observed for workers most heavily exposed and employed for 15 years or more. Dose-response for degree of arsenic exposure was observed in every duration of exposure category. Lee and Fraumeni stated that the eight-fold excess in lung cancer mortality observed among heavily exposed workers could not be accounted for by an unusually high prevalence of smoking. They cited as support for this conclusion the fact that age-adjusted death rates GTR 000114 25 for lung cancer among U.S. males for 1960-63 indicated a maximum 2- to 2-1/2-fold excess in rates for heavy smokers compared to the general population. The results of this study, although suggestive of an asso ciation between exposure to arsenic and development of lung cancer, may be challenged on the basis of lack of cigarette smoking information and presence in the work environment of other chemicals, especially sulfur dioxide, a lung irritant that may have an impact on lung cancer. Ott et al. (59) reconstructed a cohort of 603 male workers who had been engaged in the production of arsenic-containing pesticides. The 1940-73 mortality experience of these workers was determined for 10-year age groups and 5-year calendar periods and compared with expected respiratory cancer mortality derived from the corresponding U.S. white male population. Results from this retrospective cohort analysis revealed an SMR of 3.45, which represents an almost 3-1/2-fold increase in observed over expected lung cancer mortality for workers with arsenic exposure. A cross-sectional review of smoking habits of workers exposed to arsenic carried out in 1968-72 indicated no differences in cigarette consumption related to degree of exposure. Ott and his co-investigators concluded that the role of arsenic in respiratory .cancer development may be causal; however, it should be noted that their analysis is based on a small number of observed respiratory cancer cases (N=20). Milham (60) conducted a retrospective cohort study of workers ' employed in a copper smelter that produced arsenic trioxide. The workers were followed for the period 1950-71, and 40 lung cancer deaths were detected, compared to 18 expected. The SMR was 222, indicating a 2-fold excess in lung cancer mortality for the smelter workers. GTR 000115 26 As previously mentioned, there have been 11 studies of various ^\'U?S of wcrkers exPosed to arsenicals. Nine of these studies have plicated an association between lung cancer mortality and exposure to <enic compounds. The studies by Ott et al- and Lee and Fraumeni o' ' ^,,-mstrated a dose response relationship between exposure and subses*nt lung cancer mortality. A retrospective cohort study was recently conducted to evaluate the ^vrtality experience of a group of lead smelter and battery plant ^rators (61). Seven thousand thirty-two workers who had been employed least one year in the lead production industry were identified, ^ their mortality experience from 1946 through 1970 was ascertained. cancer SMR's, with expected numbers based on the U.S. male popu'*tion, were 14S for smelter workers and 132 for battery plant opera>v.rs. In addition to lead these workers were exposed to arsenic, cad- and sulfur dioxide. Again, the complexity of the work environment gjtXes it difficult to assess the effects of any one agent on lung cancer the workers exposed. It has been demonstrated experimentally that ferric oxide enhances the response to carcinogens in vivo: Ferric oxide (Fe203) is not itself * carcinogen. Rather, it acts as a vehicle for an absorbed carcinogen jai increases retention of the carcinogen in the respiratory passages of experimental animals (62) . Iron ore miners are exposed to F e ^ , and there is some indication that they may have an elevated lung cancer xrtality risk. Boyd et al. (63) examined 5,811 death certificates of British hematite (iron ore) miners. The deaths included in the study .spurred between 1948 and 1967. The.proportional mortality ratio (PMR) tvxr lung cancer vas 174 for underground miners, based on 36 observed GTR 0 0 0 1 1 fi 27 lung cancer deaths. This represents a 74% excess in the proportion of lung cancer deaths among all deaths for the miners. Radiation, which is known to have carcinogenic activity, was also present in the under ground mine environment. Thus, the observed excess of lung cancer deaths cannot be ascribed only to exposure to iron ore dusts on the basis of findings presented in this study. A causal association between occupational exposure to asbestos fibers and lung cancer and pleural mesothelioma has been well demon strated (64,65). Selikoff (65) reported the lung cancer mortality experience of 17,800 asbestos insulation workers for the period 1967-72. He found a nearly 5-fold excess in observed deaths compared to expected deaths calculated on the basis of U.S. age-specific white male death rate data. The greatest excess (7-fold) was observed among workers having 30-34 years between year of first exposure and year of death. In another study, Selikoff and his co-workers (64) followed 370 asbestos workers with a minimum of twenty years of exposure to asbestos for 52 months (1963-67). Taking cigarette smoking habits and age into account, they found that asbestos workers' risk of dying from lung cancer was 7.6 times higher than expected. When asbestos workers who also smoked were compared to non-smokers not having previous exposure to asbestos, a 92 times higher than expected risk was estimated, suggesting multiplication of the effects of cigarette smoking and asbestos exposure. Nicholson (65) studied the 1959-71 mortality of 689 asbestos production and textile workers. The study population consisted of workers who were current employees in 1959 and who had first been hired before 1939. The SMR, based on 27-deaths, for this group was 333. GTR 000117 <A< -V 28 Talc is a mineral that contains, in varying amounts, both free silica, which is a cause of pulmonary fibrosis, and different forms of asbestos. There have been conflicting reports in the literature regarding the possibility of increased lung cancer risk among workers engaged in talc mining and milling, as discussed below. Kleinfeld, et al.(66), identified 260 talc workers who were current employees in 1940 and had worked for at least 15 years or who accumu lated this amount of work experience between 1940 and 1969. Twelve cancers of the lung were observed during the follow-up period, compared to 3.7 expected on the basis of the proportion of all deaths due to lung cancer for the U.S. white male population in 1955. Thus, the Proportional Mortality Ratio of 3.2 indicated that the proportion of deaths due to lung cancer was slightly more than three times higher than expected for talc workers. In contrast, Rubino, et al.(67) found no excess in lung cancer among Italian talc miners and millers. One thousand five hundred fourteen miners and 478 millers who began work in 1921 through 1950 and who had at least one year of work experience involving exposure to talc were.studied. Follow-up for this group was conducted through June, 1974. It is difficult to compare the findings of these two studies because of differences in minimum duration of employment criteria. The talc workers studied by Kleinfeld et al.were probably more heavily exposed, since they had to have worked for at least 15 years to have been included in the study. Furthermore, the type of talc to which workers were exposed was not the same for both studies. While the talc workers studied by Kleinfeld et al.were exposed to industrial talc con taining relatively large amounts of free silica and asbestos, the GTR 000118 29 Italian miners and millers worked with very pure talc used in pharma ceutical and cosmetic industries. Thus, differences in study design and in the nature of qualitative and quantitative exposures may account for the inconsistent findings with respect to lung cancer mortality among talc workers. Radiation is considered a physical, rather than chemical, carcino gen. The association between radon daughter exposure and subsequent respiratory cancer mortality was examined in a cohort study of 3,366 white male Uranium miners, conducted between 1957 and 1968 (68). Cumu lative exposure to radiation was- measured in terms of Working Level Month (WLM), a unit reflecting both intensity and duration of exposure. Excesses in respiratory cancer mortality for Uranium miners compared to white males of the same geographical region were detected for every cumulative exposure level higher than 120 WLM. Further examination of a subset of these miners, followed prospectively from 1964 through 1967, revealed respiratory cancer mortality rates among smoking miners that were approximately nine times the rates observed in a non-mining population with similar smoking habits during the same time period (69). Workers occupationally exposed to certain organic compounds have been shown to experience an elevated risk for lung cancer. Among these high risk industries are the following: poison gas producers, ion exchange resin manufacturers, synthetic rubber and plastic producers, coal carbonization workers, and roofers. Epidemiologic studies exam ining the association between employment experience in these occupations and lung cancer mortality are summarized in Table 2-b, and will be reviewed below. GTR 000119 TABLE 2-b OCCUPATIONAL LUNG CANCER: INDUSTRIAL EXPOSURES TO ORGANIC COMPOUNDS Industry/Occupation Suspected Agents Experimental Data* Estimated Magnitude of Excess Poison Gas Producers; Military Mustard g-s (33^dichlorodiethyl sulfide) ........ . . .+ SMR = 3667 Reference 8 . 70 Ion Exchange Resin Manufacturers B C M E ............ . . + CIR^ = 8 72,73 Synthetic Rubber & Vinyl chloride . . . . . . + SMR = 156 74 Plastics Production SMR = 135 75 Bituminous Coal ' Polycyclic aromatic Carbonization Workers: hydrocarbons (coal tars).............. + Coke oven workers gas producers (retort workers) SMR = 252 (total coke oven} 76 SMR = 1000 (coke oven, topside for 5+ years 76 SMR = 3000 (Japanese gas workers) 79 SMR 180 77,78 * A "+" indicates that results of experimental tests are positive for agent of interest; a indicates negative results. i Cumulative incidence ratio. 31 Wada et al. (70) identified 2,620 former employees of a poison gas factory operated between 1925 and 1945 in Japan. Four hundred ninety-five of these workers were known to have manufactured mustard gas, 33'-dichlorodiethyl sulfide. Deaths for the years 1952-67 were ascer tained, and 33 deaths from respiratory cancer were observed among exposed employees. The expected number for 1952-67, based on Japanese national mortality rates for males, was 0.9. Thus, the SMR of 3,667 indicated a very large excess in lung cancer mortality for workers who were exposed to mustard gas. Not all former employees were included in the cohort of exposed workers, since it was estimated that approximately 5000 workers had been employed in the past at the plant. However, it is unlikely that the lung cancer mortality experience of the excluded workers was sufficiently lower than that of the workers included in the study to reverse the conclusion derived from the study that exposure to mustard gas was responsible for the excessive lung cancer mortality observed. It should be noted that the workers were exposed to other gases that are strong pulmonary irritants. Workers involved in the manufacture of ion exchange resins consti tute the principal group exposed to chloromethyl-methyl ether (CMME) and its frequent contaminant, bis-chloromethyl ether (BCME). BCME has been shown experimentally to be a potent pulmonary carcinogen (71). There is also epidemiologic evidence indicating that BCME is a strong lung carcinogen in workers exposed to this chemical. Figueroa et al, (72) followed 125 chemical plant workers, some of whom were exposed to BCME, between 1962 and 1972. Four lung cancer cases occurred during the first five years of follow-up, resulting in a five-year incidence of 4.54%. This result was compared to the incidence of lung cancer in a GTR 000121 32 referent group having a similar age distribution and not exposed to CMME or BCME. The five-year incidence for the latter group was 0.57%. The ratio of the incidence in the exposed to that in the unexposed is 9.24, indicating a 9-fold excess in lung cancer for the exposed workers. The 10-year follow-up for the same group of workers confirmed a high lung cancer incidence among those exposed to BCME (73)- For the entire 10 year period, there were 16 deaths; 11 of these were due to lung cancer; and all lung cancer deaths occurred among men who were less than 55 years of age. Furthermore, the smoking habits of the study subjects were evaluated. An inverse relationship between amount of tobacco smoked and lung cancer incidence was observed. Since the authors made no attempt to identify and trace workers who may have been exposed to BCME and who may have terminated their employment prior to the initiation of the study, the possibility that selective transfer out of the industry by heavy smokers who were exposed to BCME may have occurred cannot be ruled out. Such selection may have influenced the results of the study. A retrospective cohort study by Waxweiler et al, (74) examined the mortality experience of 1,294 workers who had been engaged in the polymerization of vinyl chloride. The lung cancer SMR reported was 156, based on 12 observed lung cancer deaths. -Tabershaw and Gaffey (75) reported an SMR of 135 for a subset of highly exposed vinyl chloride workers. The results of these studies suggest that vinyl chloride exposure may be associated with an increased risk for lung cancer. Excesses in respiratory cancer have been reported for coal carbonization workers: specifically, for coke oven workers in the U.S. steel industry (76), for retort workers in Great Britain (77,78), and for gas generator workers in Japan (79) . GTR 000122 34 Kawal et al, (79) carried out an investigation of the mortality experienced by gas generator workers at a Japanese steel plant. The cohort of interest had worked at the plant until it was closed in 1953 and consisted of 504 male workers. 25,760 other steel workers comprised the comparison group. During the follow-up period (1953-65), six lung cancer deaths were detected, compared to 0.18 expected. It is difficult to make conclusions regarding the strength of the association between exposure to coal carbonization products and lung cancer, based on the findings of this study, because of the small numbers of observed and expected deaths. However, the results of Lloyd's study and Doll's studies, in addition to those reported by Kawai et al., appear to give ample support to the hypothesis that exposure to polycyclic aromatic hydrocarbons produced during the coal carbonization process increases lung cancer risks. C. Lung Cancer Findings in the Rubber Industry Occupational exposures to hazardous respirable organic and inorganic particulates have been shown to increase lung cancer mortality among exposed populations. Substances that may be potentially harmful to the respiratory system are or have been encountered in the occupational environment in the rubber industry. These materials have at some times included known lung carcinogens, such as lead chromate, which is used in pigments, and cadmium compounds, which have been used as accelerators in the vulcanizing process; lung irritants, such as carbon black, zinc oxides, and' talc (which may contain varying amounts of asbestos - a known carcinogen); and tumor promoters, such as phenols. GTR 000124 35 Several early studies of occupational mortality in the U.S.. reported findings that include considerations of the rubber industry. In 1961-63, a national study of cause-specific mortality in 1950 and occupation as recorded on death certificates noted an excess of respi ratory cancer for employees of the rubber industry (80) . In 1967, a study of disability insurance benefits awards made between 1959 and 1962 to male workers under the age of 65 reported an excess in respira tory cancers for rubber industry employees (81). Since then, several studies of mortality in the British and U.S. rubber industries have been reported in the literature. Table 3 provides a description of the populations of rubber workers studied and a summary of the major findings, relevant to lung cancer, of these investigations. Fox et al. (4,5) examined the mortality experience of 40,867 rubber workers in Great Britain between 1967 and 1971 and 1972-74. The study population was defined as all male workers at 381 firms identified by a 1967 census of the British rubber industry. All workers included in the study had to have had at least one year of employment at the beginning of the study. SMR's were computed using annual age- and cause-specific rates of the male population of England and Wales. For 1967-71, the overall lung cancer SMR was 119. For the tire manufactur ing sector of the industry the SMR was 134, and the SMR for the sector designated as "Belting, hose rubber with asbestos, flooring," was 164; Lung cancer SMR's for the period of 1972 through 1974 were 118 for the overall industry, 128 for the tire manufacturing sector, and 106 for belting, hose rubber, and flooring. Thus, there was an excess in lung cancer mortality for the industry as a whole and for the tire sector for both follow-up periods. " " GTR 000125 ) ) ) TABLE 3 LUNG CANCER MORTALITY IN THE U.S. AND BRITISH RUBBER INDUSTRIES: OVERALL FINDINGS . i InvestigatorsT Period of Observation Fox, et al. (1974)(4) Total Industry .......... ..... Tire Manufacturing...... . Belting, Hose Rubber, and Flooring............ . 1967-71 11 11 N of Workers Studied 40,867 16,035 4,350 Standard Population 1967-70 male population of England and Wales 11 11 Estimated Effect # SMR=119 (304) SMR=134 (131) SMR=164 (42) Fox 5 Collier (1976)(5) Total......... 1.. 1...... ..... Tire Manufacturing...... Belting, Hose Rubber, and Flooring............ 1972-74 11 11 40,867 16,035 4,350 1967-73 male population of England 8 Wales 11 It SMR=118 (281) SMR=128 (117) SMR=106 (25) McMichael, et al. (1974)(82) Akron Plant Plant Plant Plant................. , 1 9 6 4 - 7 2 B ................. .. 11 C ............. ........ . It D ......... ............ 11 6,678 108* 265* 189* 1968 U.S. male population tl 11 . 11 SMR= 84 SPMR=169 SPMR=153 SPMR=229 (91) (11) (17) (37) Andjelkovich, et al.(1976)(83)......1964-73 8,418 1968 U.S. white male population Monson, et al. (1976)(8) Retrospective Cohort....... ...1940-74 13,571 1940-69 U.S. Rubber Worker Deaths, i Proportional Mortality..... ...1925-39 688* 1925-39 U.S. * References are noted in parentheses, following date of study publication. It The number of observed lung cancer deaths is given in parentheses. * Deaths, only, were considered white males white males SMR= 80 (120) SMR= 92 (234) SPMR=14Q (6) W On 37 In contrast, McMichael et al. (82), in a retrospective cohort study of 6,678 U.S. male rubber workers followed from 1964 through 1972, found no excess in respiratory cancer mortality for the industry as a whole. In a proportional mortality study of five small plants, an excess in the proportion of lung cancer deaths among all deaths was found for three of the plants. Similarly, Andjelkovic et al, (83), using a retrospective cohort analysis, studied the 1964-73 mortality experience of a separate group of 8,418 white male rubber workers. The latter investigators reported an SMR of 80 for cancer of the respir atory system among all workers, ages 40-84, indicating an overall deficit in lung cancer deaths for the rubber workers compared tc the 1968 U.S. general population. However, the investigators emphasized that use of broad age ranges in computing SMR's can mask excesses in respiratory cancer mortality found for some age groups. As an example, they cited a respiratory cancer SMR of 148 found for workers aged 40-49. Both of the preceding two cohort studies included in their separate cohorts all workers, aged 40-84, who were active or retired living employees of the rubber industry plant concerned as of January 1, 1964. Most cohort members had accrued 10 or more years- of rubber industry employment prior to definitions of the cohorts. Monson et al. (8) reported the 1940-74 mortality experience of a cohort of 13,571 white male rubber workers. They found an overall lung cancer SMR of 92, again indicating no excess lung cancer mortality for the industry as a whole. Given data restrictions regarding the quality and availability of information, the possibility of increased risk for lung cancer asso ciated with rubber industry employment can be explored on two levels. r,TR 000127 38 First, risk for the rubber industry as a whole, considering the experience of all workers combined, regardless of their particular job experience within the industry, can be examined. This approach was used in the previously described studies, and the resulting findings were, with one exception, not indicative of an excess risk of lung cancer for all rubber workers. It has been suggested that considera tion of the experience of rubber workers as a whole may lead to an inadequate assessment of risk for certain groups of workers, since exposures to hazardous materials are not uniform for all workers. Thus, an alternative approach, consisting of an evaluation of the risk asso ciated with employment in specific work areas, must be used. Several studies, already mentioned above, have attempted to do this. .Major findings, relevant to lung cancer, with special reference to specific work areas or jobs, are summarized in Table 4. Fox et al. (4,5) found elevated SMR's for lung cancer for workers engaged in tire curing and finished goods, stores and packaging dis patch. The SMR for curing was 179 for the 1967-71 follow-up period and 111 for the 1972-74 period, while the SMR for finished goods, stores, and packaging dispatch was 193 and 129 for the first and second follow-up periods, respectively. Mancuso (6) attempted to identify, through Social Security Adminis tration records, all employees at a rubber products manufacturing plant in 1938 and 1939. Data was gathered on 1,568 white male workers, and their 1940-64 mortality experience was ascertained retrospectively. Mancuso found that age-adjusted respiratory cancer mortality rates for production workers in the compounding, mill, and calender department were twice the rates for Ohio males for ages 25-64 and 5.6 times the rates of Ohio males for ages 65 and above. AnA1 Ao TABLE 4 RESPIRATORY CANCER MORTALITY IN THE U.S. $ BRITISH RUBBER INDUSTRIES: RELATIONSHIP TO SPECIFIC JOBS '" ' Investigator(s) Type of Study Mancuso, et al. (1968) Retrospective (6) Cohort Work Areas/Jobs of interest Effect Measures Compounding Milling............ Rate Ratio = 2.1 (ages 25-64) " " ............ " ' = 5 . 6 (ages 65+) Mancuso (1974) (7) Retrospective Cohort Compounding Milling............ Rate Ratio = 0.57 (Company 2) " " ............ " " = 0 . 5 4 (Company 3) Tire Tube Building ............ Rate Ratio = 1.7 " " " ............. Rate Ratio = 1.3 (Company 3) (Company 4) Curing............................ Rate Ratio = 1.8 " Rate Ratio = 1.9 " Rate Ratio = 5.4 (Company 2) (Company 3) (Company 4) Fox, et al. (1974) (4) Retrospective Cohort Curing......................... ,..SMR = 179 Finished Goods, Stores, Packaging Despatch.......... .,..SMR = 193 Fox Collier (1976) (5) Retrospective Cohort Curing..................... ... ...SMR = 111 Finished Goods, Stores, $ Packaging Despatch........... ...SMR = 129 * References are noted in parentheses, following date of study publication. T A B L E 4, c o n t i n u e d Investigator(s)* Type of Study Work Areas/Jobs of interest Effect Measures McMichael, et al., (1976)(9) McMichael, etal., (1976)(10) * Monson, et al., (1976)(8) Case-Control: . N of Cases=61 N of Controls=61 "Hybrid" N of cases=119 N of controls=1482 (based on a 22% sample of the historical cohort) Retrospective Cohort Curing.................. Tire Building........... .... Odds Ratio = 1. 66 (based on exposure rate for cases=0.25 (15/61); exposure rate for controls=0.15 (9/61)) .... no association observed Receiving S Shipping.... Compounding 5 Mixing.'.... .... RR Mill Mixing............. .... RR Extrusion............... .... RR Reclaim................. .... RR Tire Building........... .... RR Curing.................. .... RR = 1.9 = 1.4 = 2.1 = 1.4 = 2.3 = 0.6 = 0.8 Curing.................. .... SMR - 160 Andjelkovich, et al., Retrospect ive (1976)(84) Cohort ' Synthetic Latex Manufacturing........... .... SMR = 434 * References are noted in parentheses, following date of study publication. 42 McMichael et al. (10) carried out a case-control study of the association between specific rubber industry jobs and various causes of mortality. One hundred nineteen lung cancer deaths that occurred among rubber workers in a defined cohort between 1964 and 1972 were defined as cases, and a 22% sample (1,482 workers) of the original, intact cohort served as a comparison group. Rates of exposure to various work areas or jobs, directly age-adjusted to the age distri bution of the sample, were computed. The exposure rates for cases and controls were compared by constructing rate ratios for each work area. A rate ratio of 1.0 signifies that there is no difference between cases and controls in exposure to a given work area. Lung cancer rate ratios of 1.9, 1.4, 2.1, 1.4, and 2.3 were observed for Receiving and Shipping, Compounding and Mixing, Mill Mixing, Extrusion, and Reclaim, respectively. Andjelkovich et al.(84) analyzed the relationship between work experience and mortality due to specific causes for the cohort of rubber workers mentioned above. Each worker was assigned a "most representa tive department"-- defined as the department in which the^ worker spent the longest amount of employment time. A system of classification whereby rubber industry departments and work areas are grouped, accord ing to production process similarities, into broader categories called "occupational title groups" (OTG's) was applied to the observed most representative department designations to obtain "most representative OTG's'* (MROTG's). SMR's were calculated for each MROTG in order to identify work areas associated with excess lung cancer mortality. Expected numbers of deaths for each MROTG were calculated using the entire cohort as the referent group. A statistically significant excess r'-rrt n n n m 43 (SMR=434) in lung cancer deaths was observed for only one MROTG-- Synthetic Latex Manufacturing. However, associations between lung cancer mortality and exposures occurring in other work areas may not have been detected in the above study, since workers' experience in work areas not designated as their MROTG's could have had an effect on lung cancer mortality patterns. In summarization, four independent studies reporting results for lung cancer mortality among rubber workers in relation to specific jobs have given some evidence that employment in jobs involving the curing process may be associated with excess lung cancer mortality, and two studies have reported positive findings of this nature for the com pounding and mixing process. Additional work areas have been impli cated, but results have been inconsistent. Explanations for this inconsistency remain unclear. However, differences in study design, non-comparability across studies of work areas considered, and the possibility that observed associations are due to chance may contribute to variation in study results. Explanations cannot presently be advanced for the apparent deficit in lung cancer mortality for the U.S. rubber industry as a whole, in ... contrast with the excess reported for the British rubber industry. However, it is possible that voluntary exclusion from the U.S. rubber industry of the individuals who are more susceptible to lung cancer has occurred. This selection process and the effects it might have on findings with regard to lung cancer mortality in the rubber ' worker populations studied will be discussed below in general terms. Observations from occupational mortality studies indicate that working populations have a more favorable mortality experience, in GTR 000133 44 erms of overall mortality, than the general populations to which they are frequently compared. It has been suggested that this phenomenon is the result of a selection process commonly referred to as the "healthy worker effect" (85). Such an effect may occur when one or all of the following processes acts upon the working population of interest: (1) failure of individuals having manifest disease to seek employment; (2) exclusion, according to industry pre-employment medical screening policies, of the less healthy candidate employees from the active work force; and (3) voluntary self-selection out of the work force by less healthy employees, who may have greater susceptibility to the effects of industrial exposures. Fox and Collier (86) have referred to mortality-related consequences of the first and second factors as selection effects and to those of the third factor mentioned as sur vival effects. The impact of the healthy worker effect is frequently reflected in the magnitude of the all-causes SMR for the working population compared to a general referent population. An overall age-race-sex-adjusted SMR for a population of workers may be observed that is less than the SMR of 100-- expected if the force of mortality for the industrial popula tion were the same as that for the referent population--and is so often within the range of 60-80. The process of self-selection of workers out of the industry affects the calculation of SMR's only when these former employees of an industry are not included in the population-atrisk and followed for ascertainment of their mortality experience. If follow-up for all workers ever employed is achieved, the health worker effect has its greatest impact through the exclusion of candidate employees having diseases or, perhaps, risk factors for diseases that GTR 000134 45 are manifest at the time of pre-employment screening. With respect to a chronic disease whose symptoms are more likely to be detectable in older than in younger individuals, pre-employment screening will not result in effective exclusion of young workers destined to develop the chronic disease. For lung cancer, the impact of pre-employment screening on reducing the number of individuals in the work force who are at high risk for this disease is difficult to estimate. Smoking may be con sidered a risk factor for several diseases. If a potential hiree has this habit or a disease whose etiology includes exposure to cigarette smoking, he may be excluded on this basis. To the extent that smoking is a strong determinant of lung cancer and smokers are excluded from the work force, the numbers of high risk workers may be reduced. This explanation for the low observed lung cancer SMR's among rubber workers is purely speculative. It is not known to what extent smokers--and especially heavy smokers-- are excluded from entering the work force and, thus, to what extent the healthy worker selection effect is opera tive in the rubber industry. The cohort studies of rubber workers' mortality conducted by McMichael et al. and by Andjelkovich et al. excluded from consideration most workers who were employed in the industry for fewer than 10 years. It is possible that those workers who were employed in the rubber industry for comparatively short periods of time were at higher risk for lung cancer than those who remained in the work force. The exclu sion of short-term workers from enumeration in the working populationat-risk may thus have contributed to lower than expected SMR's. Fox and Collier (86) have recently studied the effects of short-term versus GTR 000135 long-term employment in the work force among a cohort of workers exposed to vinyl chloride monomer. They analyzed separately the mor tality experience of the following two groups of workers: 1) short-term workers who left the work force before having accrued 15 years of employment; and (2) workers who had at least 15 years of cumulative experience in the industry. Sixty percent of the first group had worked for fewer than five years. Fox and Collier found that the lung cancer SMR was 156 for short-term workers, while for workers continuing to be employed, it was 50. This difference was ascribed to the reten tion in the active work force of healthier individuals. D. Histologic Types of Lung Cancer Five major histologic types of lung cancer have been described by the World Health Organization (WHO) classification system, introduced in 1967 (87): 1) Epidermoid, also called squamous, carcinoma, characterized by central or proximal site of origin and by metaplasia of , columnar epithelium to form a stratified squamous epithelium with keratinization and intercellular bridges. 2) Small-cell carcinoma, including oat cell carcinoma, characterized by central location and small, round of oval cells, sparse cytoplasm, prominent nuclei, and lack of stratification. 3) Adenocarcinoma, characterized by peripheral or distal loca tion and by formation of gland-like structures and intra cellular mucin. 4) Large-cell undifferentiated carcinoma, characterized by the presence of tumor cells that are very large, are sometimes multinucleated, and have vesicular nuclei. 5. Mixed carcinomas, which usually consist of epidermoid and adenocarcinoma or oat-cell combined with other differ entiated types. GTR 000136 TABLE S HISTOLOGIC TYPES OF LUNG CANCER: ESTIMATES OF RELATIVE FREQUENCIES OF FOUR MAJOR TYPES investigator(s) Referenced Vesner, et al. 92 Shinton 89 Berg . , 94 Weiss, et al. 96 * Oat-cell carcinomas, only. Epidermoid (Squamous) 49.1% 57.2% 47.0% 61.3% Relative Frequency Small-cell Undifferentiated Adenocarcinoma 19.2% 30.2% 17.0% 12.9% 15.8% 4.0% 10.0% 22.6% Large-cell Carcinoma 15.5% - 17.0% 6.5% 49 will be more frequent in autopsy series. Thus, in series not based on complete ascertainment of lung cancer cases in a defined population, the distribution of histologic types will be subject to selection bias. The problem of selection bias may be alleviated by use of a cohort study design. Weiss, et al-(96) in a prospective study of 6,136 men at least 45 years of age at the beginning of the study, detected 121 cases of lung cancer during the 10-year follow-up period. Subsequent analysis was confined to 2,580 men who were current smokers at the initiation of the study, since lung cancers developed only among this subset of the original study population. The following distribution of histologic types based on material available for 67 of the cases, was reported: epidermoid, 61.3%; small-cell, 12.9%; adenocarcinoma, 22.6%; and large-cell, 6.5%. Although this study design would, in theory, diminish the possibility of selection bias, histologic diagnosis was unavailable for almost half the cases. Therefore, because of incomplete ascertainment of histologic types, it cannot be assumed that the ob served pattern was an accurate reflection of the true distribution of histologic types. Shinton (89) pointed out that each histologic type of lung cancer has characteristic biological features and cited differences among the various types in age and sex distribution, location in the respiratory system, occurrence of metastases, and survival prognosis. He concluded that lung cancer is a heterogeneous disease and that each of the histologic types may be associated with a distinct etiology. Cigarette smoke was the first etiologic agent to be examined in terms of its influence on the distribution of histologic types of lung cancer. rtr nnnn<) 50 Kreyberg (88) studied a series of 522 male and 78 female lung cancer cases collected in Norway. He suggested that epidermoid and small-cell carcinomas are associated with cigarette smoking, while the development of adenocarcinoma is not influenced by cigarette smoke. He calculated proportional mortality ratios (PMRs) for the various types for smokers compared to non-smokers. Comparisons of the PMRs over categories of amount smoked demonstrated that PMRs for epidermoid and small-cell carcinomas increased for each 5-gram increment in amount smoked per day, while PMRs for adenocarcinoma remained constant, over all smoking strata. In addition, Kreyberg noted that adenocarcinoma accounted for 45% of the 78 female cases but only 13% of the male cases. He postulated that this observed difference was a product of lower prevalence of smoking among females and consequent elevated frequency among females compared to males of adenocarcinomas, possibly developing in response to etiologic factors other than cigarette smoke. Yesner et al .-(93) examined histologic type and smoking habits of 449 lung cancer cases derived from a biopsy series collected between 1953 and 1959. Adenocarcinoma, with a frequency percent of 57.1%,-was the predominant type among non-smoking cases, whil-e among cases who smoked, epidermoid carcinomas predominated. Auerbach et al.(97) examined an autopsy series that included six non-smoking cases, all of whom were diagnosed as adenocarcinoma. Age-adjusted percentage distri butions of the histologic types found among smokers were examined by cigarette smoking categories. The percentage of adenocarcinomas decreased from 29.2% among those smoking less than^one pack per day to 20.1% among those smoking two or more packs per day. In contrast, the per centage of small-cell carcinomas increased from 19,2% for those smoking n n n iAn 50 Kreyberg (88) studied a series of 522 male and 78 female lung cancer cases collected in Norway. He suggested that epidermoid and small-cell carcinomas are associated with cigarette smoking, while the development of adenocarcinoma is not influenced by cigarette smoke. He calculated proportional mortality ratios (PMRs) for the various types for smokers compared to non-smokers. Comparisons of the PMRs over categories of amount smoked demonstrated that PMRs for epidermoid and small-cell carcinomas increased for each 5-gram increment in amount smoked per day, while PMRs for adenocarcinoma remained constant, over all smoking strata. In addition, Kreyberg noted that adenocarcinoma accounted for 45% of the 78 female cases but only 13% of the male cases. He postulated that this observed difference was a product of lower prevalence of smoking among females and consequent elevated frequency among females compared to males of adenocarcinomas, possibly developing in response to etiologic factors other than cigarette smoke. Yesner et al .-(93) examined histologic type and smoking habits of 449 lung cancer cases derived from a biopsy series collected between 1953 and 1959. Adenocarcinoma, with a frequency percent of 57.1%,-was the predominant type among non-smoking cases, whil-e among cases who smoked, epidermoid carcinomas predominated. Auerbach et al. (97) examined an autopsy series that included six non-smoking cases, all of whom were diagnosed as adenocarcinoma. Age-adjusted percentage distri butions of the histologic types found among smokers were examined by cigarette smoking categories. The percentage of adenocarcinomas decreased from 29.2% among those smoking less than^one pack per day to 20.1% among those smoking two or more packs per day. In contrast, the per centage of small-cell carcinomas increased from 19.2% for those smoking r''rD n n n i a n 51 less than one pack per day to 31.1% for those smoking 2+ packs per day. No differences according to amount smoked were noted for epidermoid or large-cell carcinomas over the same range of cigarette consumption. In contrast to the above studies, which suggest an association between smoking and distribution of histologic type of lung cancer, the study of Weiss et al.(96) does not indicate that such a relationship exists. The latter investigators compared age- and race-adjusted rates of mortality due to the five major types of lung cancer among categories of smoking and found that epidermoid, small-cell, and adenocarcinomas all showed a dose-response relationship with amount of cigarette smoking. There is evidence to suggest that environmental carcinogens other than cigarette smoke may also influence the distribution of histologic types of lung cancer in exposed populations. Whitwell et al. (98) studied the histologic types of lung cancer found among workers with certified asbestosis who died in the United Kingdom between 1962 and 1972. There were 88 male cases available for study, or 67% of all lung cancer cases having concomitant asbestosis. Adenocarcinoma was found to be the predominant type of lung cancer for all grades of asbestosis, and the distribution of cigarette consumption clid not vary among the histologic types. The authors of this study concluded that their identification of adenocarcinoma as the predominant type of lung cancer among workers with asbestosis was unusual. They suggested that asbestos dust can reach the distal areas of the bronchial tree and, acting as a cocarcinogen with cigarette smoke, can give rise to adenocarcinomas. Saccomanno et al .(99) studied 121 cases of lung cancer that occurred between 1949 and 1969 among Uranium miners, in order to evaluate the association between.'radon daughter exposure and distribution GTR 000141 52 of histologic types of lung cancer. One hundred thirty-eight lung cancer cases in non-miners, occurring in the same geographical area during the same time period, constituted a referent group and were matched with miners on age and cigarette smoking habits. Distributions of histologic types among miners and the referent group were compared. At every level of radon daughter exposure, the proportion of small-cell carcinomas was higher for Uranium miners than for the referent group. The authors concluded that radiation exposure influences the distri bution of histologic type of lung cancer among Uranium miners and that , while cigarette smoke is a potent co-carcinogen, it exerts little influence on histologic type of lung cancer in the population of miners studied. ' Figueroa et al. (72) as mentioned previously, reported-the 1962 1967 lung cancer mortality experience of 125 men with industrial exposure to chloromethyl methyl ether (CMME) and it contaminant BCME. Four cases of lung cancer occurred. Ten additional cases were detected among employees exposed to CMME and BCME prior to 1962. Histologic type was determined for 13 of these cases, and 12 were classified as small cell carcinomas. Although the series of cases is small and ascertain ment of cases for workers exposed prior to 1962 may be incomplete, these findings suggest that the distribution of histologic type of lung cancer is influenced by exposure to CMME or BCME. Waxweiler et al. (74) in the retrospective cohort study of vinyl chloride workers described previously, were able to examine histologic material for eight of the 12 lung cancers that were detected. Five cases were classified as large-cell undifferentiated carcinoma and three as adenocarcinoma. The predominance -of large-cell carcinomas GTR 000142 53 and adenocarcinomas and the absence of epidermoid and small-cell carcinomas is unusual. Findings of the above-mentioned studies suggest that specificity of histologic type of lung cancer may be determined, in part, by etiologically relevant environmental exposures and that different carcinogens may contribute to the observed variability in patterns of distribution of the histologic types. Radon daughter and CMME (or BCME) exposures result in an elevated relative frequency of small-cell carcinomas; asbestos exposure, in adenocarcinoma; and vinyl chloride, in large-cell and adenocarcinomas. Study of the distribution of histologic types of lung cancer occurring among populations whose specific exposure is unknown may provide clues to or indicate the presence of unusual carcinogens in their environments. E. Summary of the Background Information and Literature Review 1. Lung cancer mortality exerts a considerable impact on the health of the general population and on the labor force, in terms of , work-years lost. The 1940-67 increase in rates of lung cancer mortality in the U.S. has occurred for all race-sex group? and has been most dramatic for white males, ages 35-44, for black males, and, since 1960, for females. 2. Lung cancer mortality rates.show considerable geographical variation, both within the United States and on a world-wide level. Environmental and host-susceptibility factors may, in combination, influence the observed variability. Known risk factors for lung cancer include the following: age, sex, race or ethnicity, urban residence, migration, smoking, and occupation. GTR 000143 54 3. Although trends in lung cancer mortality largely reflect cigarette consumption patterns, agents encountered in occupational settings may have an impact on lung carcinogenesis. A probable causal association between occupational exposures and lung carcinogenesis has been demonstrated for radon daughter, asbestos, arsenicals, and coal carbonization products exposures and has been suggested for a number of additional industrial exposures. 4. Studies of the rubber industry have shown lung cancer mortality to be elevated for certain work areas. However, these studies have not consistently indicated that elevated lung cancer mortality is specific for a particular work area. Therefore, it has not been possible to identify agents having etiologic significance. . 5. There is evidence to suggest that occupational exposures, such as radon daughter, BCME, and vinyl chloride, may cause the relative frequencies of the histologic types of lung cancer to fluctuate from the distribution found among a general population, whose usual exposure is cigarette smoke. Therefore, inspection of the distribution of histologic types of lung cancer found in an industrial population may provide insight into the nature, of the exposures responsible for lung cancer induction and may aid in distinguishing lung cancers arising in response to cigarette consumption from those developing a a result of occupational fexposure. GTR 000144 (2) As indicated in the accompanying literature review, certain groups of migrants, both foreign-born and those migrating within the U.S., have excessive lung cancer mortality, compared to native Ohio residents. However, studies of the relationship between migrant status and lung cancer mortality have not investigated the possibility that migrant status exerts its influence through the differential employment of migrant workers in hazardous occupations. Therefore, the second objective of this study wa6 to evaluate the residual effect, if any, of migrant status on lung cancer mortality among a relatively homogeneous population of rubber workers at an Ohio rubber plant. With respect to the first objective stated - that is, to determine 'jihich work areas, if any, are associated with an increased risk of lung cancer mortality - it was expected that the following relationships ' would be present for any work area found to be associated with increased lung cancer risk: (1) mean duration of employment in that work area would be greater for workers who died o 2 lung cancer than for those who remained - free of the disease until their death; . (2) relative risk of lung cancer for workers -having a history of employment ("exposed") in a given work area, compared to workers who never worked ("unexposed") in the work area, would increase with increments in a minimum duration of exposure criterion demonstrating consistency with the dose-response concept, where "dose" is defined A as duration of exposure to a specific work area and "response" is < indicated by lung cancer risk; and / GTR 000146 insure that cases and their respective controls had equal opportunity for exposure. The process of selection of cases and controls and the definition of exposure variables will be discussed in greater detail later. Rationale for selecting the case-control mode of investigation is presented below. A. Rationale Underlying the Choice of Study Design Choice of a case-control study design was based on the consideration of two issues. The first of these concerned the appropriateness of the design for fulfilling the primary research objective, which was to identify workers who may be at increased risk for lung cancer by virtue of their employment in specific rubber industry jobs, and the second concerned the feasibility of implementing the chosen design. In case-control studies, disease incidence rates and cumulative incidence rates are not known. Therefore, direct quantification of disease rate ratios and risk ratios (relative risks) for exposed subjects compared to unexposed subjects cannot be carried out. Information regarding the probability of disease given exposure can be directly obtained only from retrospective or prospective follow-up (cohort) studies. However, relative risk may be approximated in case-control studies by the odds ratio (100, 101). This parameter is estimated by dividing the ratio of exposed to unexposed cases by the ratio of exposed to unexposed controls. Thus, the odds ratio is the ratio of the exposure odds for cases compared to non-cases. It has been shown that, when the probability of disease in the source population is low (less than 20%), the difference between estimates of the risk ratio (relative risk) and the odds ratio is trivial (100, 101). Furthermore, Miettinen (10.2) has GTR 000149 60 59 shown that, when incident cases are considered, and controls are drawn from a defined population, odds ratios derived from case-control studies estimate incidence density ratios and, thus, instantaneous risk ratios, without the assumption that the disease of interest be rare. The casecontrol method, therefore, provides a means for estimating an appropriate effect measure - the odds ratio - which is useful for evaluating possible associations between a disease and exposure of interest and for assessing the magnitude of any association found. Prospective and retrospective follow-up studies, in which exposures are measured at one point in time and subsequent disease occurrence among exposed and unexposed subjects is ascertained during or after a specified interval of time, constitute alternative modes of investigation that offer the following advantages over case-control studies (101): (1) direct quantification of rate ratios or relative risks is possible, since either incidence or cumulative incidence rates may be calculated; (2) if attrition does not occur and if disease rates are ascertained for all members of the study population, selection bias will not be introduced; and (3) when the distorting effects of potential confounders are controlled and when selection bias and mieclassification-errors are not "present, follow-up studies provide more convincing evidence than case-control studies for the presence of causal relationships, since the follow-up method distinguishes disease that occurs as the result of exposure from that which occurs concurrently with exposure or prior to exposure. The problem of determining whether an observed association between disease and exposure is causal when a case-control study, in which exposure is measured retrospectively, is used can be partially overcome by careful exclusion from consideration of exposures that may be irrelevant because kte? n n oi50 of insufficient lapsed amount of time between the onset of exposure and disease occurrence. An additional advantage of the prospective follow up method is the possibility for investigator control over the types and accuracy of measurements of both the exposure variable(s) and potential confounders of interest. Retrospective follow-up and case-control studies often suffer from absence of useful data or inadequacy of certain aspects of available data. Major obstacles confronting the application of prospective and retrospective follow-up study designs derive from the issues of convenience and economy. Consideration of these issues was the second factor leading to the choice of the case-control study method. The prospective approach was not feasible for studying the association between rubber workers' lung cancer mortality and job exposures, since the lung cancer latent period associated with occupational exposures often exceeds 15-20 years. Therefore, the minimum follow-up period that must be allowed for the detection of any effect of exposure on risk of lung cancer precludes the use of a prospective design. In addition, since lung cancer is a comparatively rare disease, a prohibitively long amount of time or large population must be provided in order to compile a case series of appropriate size for estimating effect measures and drawing inferences with respect to these measures. Finally, subjects included in a prospective study must be classified according to exposure at the outset of the study, and this design is therefore not feasible for the investigation of multiple exposures. GTR 000151 63 or a representative sample thereof, for whom ascertainment was not determined by exposure status, are considered; (2) the controls included in the study population have exposure rates that are equivalent to those experienced by non-diseased individuals in the source population; and (3) no misclassification with respect to disease status has occurred. Violations of these assumptions may lead to systematic error or bias in estimation of relative risk and thus contribute to the major limitations of case-control studies. With regard to the first assumption, failure to enumerate all cases because of attrition from the source population prior to identification of the case series to be studied may result in the selective inclusion in the study of cases having non representative exposure histories. With regard to the second assumption, if the cases considered in the study are representative but the exposure rate among study controls is lower than the true exposure rate 'among non-diseased individuals in the source population, the effect measure (odds ratio) will be exaggerated. Conversely, if the exposure rate for the study controls is higher than the true rate among non-diseased members of the source population, the effect measure will be underestimated. Representativeness of cases and controls selected for this study will be discussed later. With regard to the third assumption, an attempt was made to ascertain all rubber worker lung cancer deaths occurring during a specified period of time (83). However, it is, in fact, not possible to estimate the number of undetected deaths due to lung cancer, and some of these may have been Included in the control series. Also, some of the subjects included in the case series may have been erroneously diagnosed as having lung cancer. - GTR 000153 64 C. Design Considerations 1. Description of the Source and Sample Populations Although it may be desirable to consider the theoretical source population for the current study as all workers ever employed at the rubber plant of interest, in fact, the lung cancer cases and referents comprising the study population were members of a cohort of rubber workers identified at a single rubber manufacturing plant in Ohio. This cohort has been described previously (83) Construction and follow-up of the cohort are reviewed below, and methods used to select cases and controls are discussed. The fixed cohort, henceforth referred to as the "1964 Cohort" consisted of all active or living retired hourly workers at the plant of interest who were at least 40 years old as of January 1, 1964. The 1964 Cohort was identified through examination of available company personnel and union records. The mortality experience of the entire cohort of 8,938 workers and the subset of 8,418 white male members was ascertained for the ten-year period, ending on December 31, 1973, and is summarized in Figure 1. GTR 000154 65 January 1, 1964 Fixed Cohort total N (all males) = 8,938 N white males = 8,418 December 31, 1973 - -Living' .--total N = 6,430 "white males, N = 6,045 Dead 1 ICDA #162 N = 131 .-- total N = 2,508 -- white males, N = 2,373 ---- I1------------ 1 All Other Causes N = 2,242 v CASES v CONTROLS Figure 1: Mortality experience of the 1964 Cohort of rubber workers. The vital status of Cohort members was determined through company insurance benefits records and the Bureaus of Vital Statistics of various states. Follow-up was 96.7% complete. Information pertaining to cause of death was obtained from death certificates available for approximately 98% of Cohort member deaths. - The 1964 Cohort excludes the following groups of workers who may be considered as members of the conceptual source population: (1) workers who were listed as members of the 1964 Cohort but who could not be traced; (2) former employees who died before 1964; (3) employees who were living but were under 40 years of age on January 1, 1964; and (4) GTR 000155 former, short-term, employees who quit the industry before qualifying to receive retirement benefits. As indicated above, the number of workers who could not be traced is small, and it is therefore unlikely that differences in disease rates or exposure patterns between these individuals and the successfully followed portion of the 1964 Cohort could affect the results of the current study. Of major concern to the validity of making inferences from results based on members of the 1964 Cohort are t the second, third, and fourth groups enumerated above. For the purpose of drawing inferences with respect to all employees of the plant from the results of this study, it must be assumed that exposure "rates" for these individuals were comparable to those experienced by workers included in the 1964 Cohort. The correctness of this assumption is not known. All deaths occurring among 1964 Cohort members during the 10-year period of 1964 through 1973 constituted the source for both cases and controls. White males only were included in the current study, since black male and female workers would have to have been considered separately for reasons of validity, and numbers of lung cancer deaths among the latter two groups were small. White males comprised approximately 94% of the 1964 Cohort of rubber workers. - As shown in Figure 1, 131 white male hourly workers died with a diagnosis of primary lung cancer (ICD #162) at some level of causation on their death certificates, These were designated as cases. The 2,242 white male deaths due to all causes other than malignant neoplasms of the respiratory system constituted the .comparison group from which the controls included in this study were selected. Five--hundred controls were individually matched with cases on year of birth and year of first hire at the plant under study. One to 4 controls were chosen for each RTR 000156 67 case, depending on availability within the matching restrictions. A detailed discussion of the selection of controls for the present'study will be presented in the following two sections. 2. Validity with Respect to Selection of Cases and Controls A desirable property of any epidemiologic study design stipulates that, under the null situation of no association between the exposure(s) and health outcome of interest, study subjects should be chosen so that exposure "rates" are the same for cases and controls (in case-control studies) or so that disease incidence rates are equal among exposed and unexposed subjects (in follow-up studies). For case-control studies in which the case group consists of all cases or a representative sample of all cases, it is desirable for the referent group also to be representative, .n terms of exposure, of all non-diseased members of the source population. Two issues relevant to the attainment of this property in the present study warrant,,discussion. First, it was necessary to limit the pool of eligible controls to all deaths, other than those due to cancer of the lung, rather than to select controls from among all members of the 1964 Cohort, regardless of vital status at the termination of the follow-up period, because complete work history information was available for deceased workers only. Thus, for any work exposure that is positively associated both with lung cancer mortality and with death due to another cause or with vital status in general, differences between cases and controls with respect to the exposure will be underestimated. In effect, then, one possible ^J^asult of this aspect of the study design will be to dilute estimates of relative risk. This limitation is inherent in the design of the study n t o nnni57 69 and cancer of another site, the ability to detect differences between cases and controls in this factor will be reduced. The inclusion in the comparison group of deaths from other cancer causes also associated with the factor in question would result in spuriously elevated exposure "rates" among the controls, and, again, effect measures estimated would be biased towards the null. In order to ascertain the extent of the possible bias introduced by including deaths due to other malignant neoplasms in the control group, the distribution of deaths due to cancer among the controls may be examined, and analyses based on the total control group (other cancer deaths and non--cancer deaths, combined) and on the subset of controls comprised of non-cancer deaths only may be compared. These latter analyses have not yet been completed. 3. Validity with Respect to Control of Potentially Confounding Variables Because of the alterations in production processes in the rubber industry, jobs and associated exposures may have changed qualitatively and quantitatively over time. For exiample, workers who entered the industry in 1930 may have had the opportunity for contact with materials that were not used in the 1940's. Thus, workers who initiated their employment in the industry at different points in time may not have had equivalent opportunity for employment in the same types of jobs or may have encountered qualitatively different exposures in the same jobs because of changes occurring over time in materials used. For this reason, it was necessary to control for year of first employment. It was decided to carry out individual matching of cases with controls on year of first employment at the plant under study (+ 1 year), since GTR 000159 stratification in ensuing analyses as an alternative control procedure would be inefficient and might lead to loss of internal validity due to differences between cases and controls in year of first hire within strata of this variable. In addition to year of first hire, race, sex, and age were thought to be potential confounders of the association between lung cancer mortality and work history patterns. These variables were considered at the design stage of the study by one of the following methods: (1) subject category restriction or (2) matching. In order to be a confounder, a factor must both be associated with the exposure of interest and be a predictor or risk factor for the disease, independent of exposure. Race and sex are known predictors of lung cancer mortality. In addition, it is possible that both a worker's race and sex may have influenced his/her choice of or assignment to jobs within the rubber industry. Age may be a confounder by virtue of its known positive association with lung cancer (i.e., lung cancer death rates increase with age) and its possible association with work history. An association between work history and age may have arisen in the study population if some factor, such as company employment practices, led to the differential assignment of workers to particular jobs depending on age. . The potentially confounding effects of race and sex were controlled for in the current study by restricting eligibility for inclusion in the study to white males. The effects of age were considered in the study design by matching cases and controls on year of birth (+ 3 years), and by requiring that controls must have lived at least as long as their respective cases. By stipulating that the controls must have achieved the same or a greater attained age at death as their respective cases, 71 it was intended to accomplish the following: (1) the possibility of systematically excluding controls who may have lived longer than respective cases because of differences in exposures will be reduced, and the ability to detect differences between cases and controls in exposures having health effects of interest will be enhanced; (2) the possibility of including as a control an individual whose exposure may have been etiologically relevant to lung cancer but who died of another cause before lung cancer became manifest will be minimized; and (3) comparisons of "old" cases with "young" controls will be avoided, and the variability of unmeasured or unknown lung cancer risk factors that are correlates of age will be controlled. The matching scheme employed in the current study - i.e., matching on both year of birth and year of first hire in the rubber industry - in addition effectively limits case-control differences in age at first hire. For example, it would be undesirable to compare a case who was first hired at age 20 with a control who was AO years of age when he initially entered the rubber industry. These individuals may differ with respect to unrecognized or unmeasured lung cancer risk factors that are correlates of age at first hire, and age at first hire may be a determinant of subsequent work history pattern: for example, older workers may be placed in less demanding jobs involving less severe . exposures and may have less opportunity for job mobility, compared to younger workers. An additional potential confounder that could not be considered because pertinent data were unavailable was cigarette consumption by study subjects. As noted previously, cigarette smoking is the strongest GTR 000161 72 known risk factor for lung cancer. If cigarette consumption were distributed unequally among the different groups of workers, categorized by their work history, who are compared in the present study, it cannot be known whether associations observed are attributable to materials encountered in the work environment or to cigarette smoking. Although it was not possible to consider the cigarette smoking habits of cases and controls, it was thought that, by matching on year of birth, cases and respective controls would at least belong to the same birth cohort of smokers. This is important since cigarette smoking habits, in general, vary according to birth cohort membership. As explained above, individual matching on year of birth and year of first hire was conducted, with the additional constraint that controls must have lived at least as long as their respective case. Controls' year of- birth was allowed to vary by plus or minus three years from their respective cases' year of birth, and year of first hire was allowed to vary by plus or minus one year. It was felt that differences between cases' and controls' values of these variables resulting from permitting ** the above ranges would be inconsequential. Furthermore, this procedure permitted the selection of more controls per case. -Four controls per case were sought in order to increase the sample size used in the study and, thus, the precision of parameter estimation. However, the matching ratio was allowed to vary from 4:1 to 1:1 when fewer than four controls were available for a case. . ' GTR 000162 D. Sources and Adequacy of DataIn order to obtain unbiased risk estimates, information with respect to the exposures of interest, disease status, and other (control) factors should be of the same quality for cases and controls. The source of data for constructing the case series and the pool of eligible controls consisted of death certificates for all deaths occurring among members of the 1964 Cohort during the ten-year follow-up period. All death certificates were nosologized according to the Eighth Revision of the International Classification of Diseases (ICD). Death certificates were not available for four (<1%) of the study subjects. For these individuals, cause of death was determined by examination of company insurance benefits records. As previously noted, It is not known to what extent study subjects were misclassifled according to their lung cancer status. Although it is possible that some of the controls selected had an undetected lung malignancy, there is no reason to believe that failure to detect lung cancer is associated with exposure (work) history. Hence, any errors due to misclassification by disease status are likely to be non differential and will lead to conservative estimates of effect. Additional information obtained from death certificates included dates of birth and death and place of birth (state or country, if* foreign-bom). Detailed job Information for each study subject was obtained from company employment records. These records will be referred to as work histories. It has been assumed that the work histories are essentially complete in terms of listing all jobs held and that the completeness and accuracy of the work histories are not dependent on case-control status, 6TR 000163 74 so that any misclassification with respect to exposure was non-differential. Again, there is no reason to suspect that differential classification errors have taken place. As previously noted, "exposure" in the current study is defined as occurrence and duration of employment in certain rubber industry production work areas. There are numerous jobs in the rubber industry, each involving diverse qualitative and quantitative exposures to chemically complex work environments, and an individual worker may move through more than 30 jobs during the course of his employment at a rubber plant (103). The chemicals that were used in or generated by a given production process can seldom be identified, and quantitative information is not available. For most jobs, both the exact chemical composition and the amount of materials constituting the major exposures are unknown. Thus, because of the chemical complexity of the work environments encountered by rubber industry employees and the currently incomplete characterization of these environments, it is not feasible in most instances to objectively classify workers according to categories of exposure to specific chemical agents and subsequently to determine disease risk associated with these exposure categories. An alternative job classification scheme - the Occupational Title Classification System - has b'een devised (103). This classification system groups together jobs on the basis of similarities in materials used, production process involved, and/or products produced. Thus, jobs that involve relatively homogeneous conditions with respect to type of raw materials or products encountered may be grouped into units called "Occupational Titles (OT's)". The resulting OT's may then be aggregated - again on the basis of products or processes involved into a smaller number of "Occupational-Title Groups (OTG's)". For the GTR 000164 75 "v present study, 119 OT's grouped into 21 OTG's were considered. OTG's 1 20 involve active employment, while OTG 21 is reserved for absence periods and exits from the industry and will not be considered in the analyses. Lists of th OT's and of the OTG's and their component OT's are given in Appendix A-l and A-2. Although a complete characterization of the OTG's is not possible, Industrial hygienists have been able to ascertain the major environmental hazards encountered in each OTG. This description is included in Appendix A-3. .. As noted previously, the present study was directed toward assessing the relationship between lung cancer mortality and employment experience in two work areas - Compounding and Mixing (OTG 1) and Curing (OTG 9) found in previous studies to be associated with increased lung cancer risk, and toward detecting possible associations with other work areas. s defined in the present study, the Compounding and Mixing work area is comprised of 12 OT's, and the Curing work area contains 7 OT's. In addition to the potential exposures mentioned in Appendix A-3, a list of specific chemicals which are reported to have been present in these OTG's, although not necessarily in this plant, and known to constitute respiratory system hazards is given in Table 6. Some of these compounds,, such as lead chromate, are or have been used in relatively small amounts. The decision to use OTG classifications rather than some other method of grouping jobs according to exposure was made for the following reasons: (1) there were no clear-cut, a priori hypotheses regarding a specific subset of jobs clearly different from the job groups composing the OTG's; (2) the OTG classification results in job groupings similar to those previously investigated by other researchers and thus facilitates in tl OTG pounding Lxing ring !irce8 : 0? Re lie suspect scussed i idmium die this Cad chroma pounds u GTR 000165 76 Table 6 Known Chemical Respiratory System Hazards Reported to Have Been Used In the Rubber Industry In the Compounding and Mixing and the Curing OTG's* l)TG founding & png Chemical Compound Cadmium ^ compounds 2 Chromates Carbon Black Zinc oxide Talc Use in the Rubber Industry -accelerator -present in pigments -filler -accelerator; activator -filler; coating of green rubber Phenols -bonding agents (phenol-form- aldehyde resins) jig Talc -see above Phenols - " Comments -Cd. is a possible carcinogen -lung irritant & carcinogen -lung irritant -lung irritant -lung irritant; may be contaminated with asbestos, a lung carcinogen -lung irritant; tumor promoter; ciliostatic agent -see above - " es: OHSG: unpublished environmental data. . . ' References 49-53, 56, 63-67, and 36. suspected roles as respiratory system hazards of mos-t of these compounds have been ussed in greater detail in the Literature Review section of this paper. um diethyldithiocarbamate is the specific accelerator used. The carcinogenicity Cadmium compound is unknown. chromate salts have been used as pigments. The carcinogenicity of the specific 8 used at the plant under study is unknown. GTR 000166 comparison of the current study*6 results with earlier reports in the literature; and (3) the use of the relatively small number of OTG's to summarize job information was necessary in order to obtain adequate numbers of subjects having employment experience in the work areas of interest for derivation of statistically stable risk estimates. The detailed work histories obtained from company records for each subject consisted of the worker's name and social security number and the following information for each consecutive job change: the complete date in and date out for the job, a description of the job, and a department code for the job. These work history records were summarized as follows: (1) the date of first employment in the rubber industry and the date on which the worker terminated the last job he held at the plant were noted; (2) the amount of time spent in each job was computed; (3) each job was assigned an OT and an OTG by a computerized job-OT-OTG dictionary; (4) the amount of time spent in each OT and in each OTG was calculated by accumulating total time spent in the jobs constituting the appropriate OT's and OTG's (if the cumulative amount of time spent in an OT was less than 15 days, the duration of employment in this OT was counted as zero); '5) the year of first employment in each OT and each OTG was computed as the earliest year in which the_-worker was employed in any job belonging to the given OT or OTG; (6) the. total amount of time spent as an active worker in the rubber plant was calculated by summing over the time spent in each OT involving active employment. Exclusions of 62 initially selected subjects occurred for the following reasons (also, see Table 8): GTR 000167 (1) if, for either cases or controls, no work history could be located; (2) if, for either cases or controls, the work history was found to be incomplete; (3) if, for controls only, the respective case was excluded because of missing or inadequate work history information; (4) if, for cases only, no suitable match was found. E. Analyses In order to assess the results and adequacy of the matching procedure, differences between each case and the average among his corresponding controls for year of first hire, year of birth, and age at first hire were computed. The mean of the differences (D) was computed for each of these variables and the null hypothesis, Hq : D = 0, was evaluated, using the paired sample t-test (see Appendix B for formulae). Also, the group means of these variables for cases and controls, ignoring the matching, were calculated for descriptive purposes. . The means for cases and controls for age at death and the mean difference between each case's age at death the average age at death among his respective controls were also evaluated in order to determine whether controls lived longer than cases and whether cases' age at death was lower than might be anticipated, since early death from lung cancer may be indicative of heavy or unusual, and possibly work-related, exposures. The frequency distribution of length of overall employment and the mean duration of overall employment were determined separately for cases and controls and were compared, using the chi-square test and the two sample t-test (see Appendix B). Also, the paired sample t-test, which GTR 000168 79 takes matching into account, was calculated for the mean difference between cases and respective controls in length of overall employment. The question of equivalency of total duration of overall employment between cases and controls is of importance in evaluating the association between OTG-specific employment and lung cancer risk. It is desirable for cases and controls to have had equal opportunity for employment in the OTG of interest. It is reasonable to expect that, as a subject's cumulative length of employment increased, his opportunity to move into a given OTG also Increased. Thus, for example, if a case had been employed for a longer period of time than his control, the case may have had a greater chance to have worked in any given OTG. If this occurred, relative risk estimates would be biased upwards. Such a bias is avoided if cases and controls are not different with respect to cumulative duration of employment or if controls having shorter duration than cases are excluded. A bias in the opposite direction may occur if cases worked for a shorter period of time than controls, and this bias may be overcome by restricting consideration of controls' work histories to the period defined by respective cases' cumulative duration of overall ; employment. Preliminary analyses were based on the complete work histories | of all cases and controls Included in the study. Migrant status for study subjects was ascertained from place of birth information recorded on death certificates. A migrant in the present study was defined as any subject not born in the State of Ohio. In order, to compare the results of the present study with those of Mancuso (40,41) and others, migrants were further classified into one of the following groups: (1) European or foreign-born migrants, (2).migrants GTR 000169 from Geor Caro migr New U.S. Mich U.S., bom grou of o base empi cont mini empi the and boti Stai coni by 1 is < eri sam OTG 80 from the southern U.S.A., including the states of Kentucky, Tennessee, Georgia, Alabama, Louisiana, Arkansas, Mississippi, Virginia, North Carolina, South Carolina, Florida, Maryland, Texas, and Oklahoma; (3) migrants from the northeastern U.S.A., including the states of Pennsylvania, New Jersey, New York, and Connecticut; and (4) migrants from central U.S.A., including the states of West Virginia, Indiana, Illinois, Nebraska, Michigan, and Kansas. Other states and geographical regions of the U.S.A. were excluded from the above classifications since no subject was born in these areas. Percentages of native Ohioans and of each migrant group were determined for cases and controls, and the statistical significance of observed differences was tested, again using the chi-square test. Next, OTG-specific employment was considered. Several analyses, based on subjects who fulfilled some minimum duration of OTG-specific employment criterion, were carried out. The percentages of cases and controls employed in each OTG were ascertained separately for three minimum duration of employment categories: (1) ever (more than 15 days) employed in the OTG of interest; (2) employed for at least 2 years in the OTG of interest; and (3) employed for 5 or more years. The second and third categories mentioned above are not mutually exclusive, -and both represent subsets of the first duration of- employment category. Statistical significance of differences between proportions of cases and controls falling into each category were tested separately for each OTG by the chi-square test. For this test, the proportion "exposed" (employed) is defined by the number of subjects meeting the minimum duration criterion in the OTG specified, divided by all other subjects having the same disease status. For example, the proportion of cases "exposed" to OTG 1 for 5 or more years is given by the number of cases whose cumulative c t o nnni 70 employment in OTG .1 .was 5 or more years in duration, divided '.by the number of cases who spent fewer than 5 years in OTG 1 or were never employed in OTG 1. It was then desired to quantify -the risk for lung cancer associated with employment in each of the 20 "active" OTG's. As previously discussed, the appropriate measure of effect, in terms of risk, for case-control data is the odds ratio :<0R). "Data layout for calculation of a -.crude odds ratio is illustrated in Figure 2- : Case-Control Status Cases Controls Total Exposed a c "1 Not Exposed b d mo 1 no N Where, a,b,c, and d represent the number of subjects falling into each exp08ure-ca.se/control-specific classification; n^=total number of cases; n^*=total number of controls; m^=total number of exposed subjects; m^total number of unexposed subjects; and U=total number of all study subjects. Figure 2. Data layout for calculation of the crude odds ratio (OR) in case-control studies. "" Source: Reference 100. The point estimate of the odds ratio is given by A ad OR be, (1) and the appropriate test of the statistical significance of the association A represented by the OR is the chi-square test with 1 degree of freedom (104) . Formulae used for computing this test statistic and for deriving 82 test-based confidence intervals are given in Appendix B. The null A hypothesis of interest is given by H^: OR = 1. /S - It has been shown (105) that the crude OR is an unbiased estimator only when the proportion of exposed who are positive for the matching variable is equal to the proportion of non-exposed subjects who are positive for the matching variable, where the matching and exposure variables are both dichotomous. This statement is applicable, also, to the current study, in which matching variables are continuous, and it implies that the crude OR will be unbiased only when there is no association between the exposure agent and the matching variate. When any such association exists, the crude OR will be biased, even when the covariate is not a confounder: i.e., even when the covariate is not associated with the disease (as is expected, as a result of the matching procedure). In case-control studies with individual matching, unbiased estimates of relative risk can be obtained only by performing stratified analyses (asstiming that exposure is associated with the matching variates), where each stratum consists of a case and his respective controls. As mentioned * previously, depletion of the pool of eligible controls in the current study resulted in there being a variable matching ratio (R). However, matched quintuplet <R=4:1) were obtained, when possible. Thus, the method used for computation of the unbiased for matched quintuples will be demonstrated; then calculation of the OR when a variable matching ratio is used will be presented. Data layout for calculation of the 6 r. given a 4:1 matching ratio is A shown in Figure 3. Referring to this figure, the OR, as derived by Mantel and Haenszel (104), is defined by the following equation: GTR 000172 83 OR ^ f10+3fll+2f12+f1 3 ^ 5 ^ ^ f04+3f03+2f02+f0 1 ^ 5 ^ Cases Exposed Not Exposed Number of Controls Positive for Exposure 4 3 2 1 0 f14 f13 f12 fll f10 V o o U-l f04 f03 f02 f01 f0" f f .4 f .3 f.2 f .l f .o Where the the number of matched groups having i cases and j controls with the exposure response indicated. Figure 3. Layout and notation for data from case-control studies with . 4:1 matching and dichotomous outcome and exposure. Sources: References 104 and 106. In the variable matching ratio situation, where the ratio may vary from 1:1 to 4:1, four subseries of matched groups, each having a fixed number of referents per case, must be considered: i.e,, pairs, triplets, quadruplets, and quintuplets. For pairs, an odds ratio is obtained as follows: ' " OR - (f10)/2 / (f01)/2 (3) For triplets, the analogous odds ratio formula is: (4) R = (2f10+fll)/3 1 <2f02+f01)/3 GTR 000173 84 and for quadruplets, the analogous formula is: OR = (3f10+2fu +f12)/4 / (3f03+2f02+f01)/4 (5) An overall point estimate of the odds ratio may be obtained by summing the numerators of Equations 2-5 over all values of R (the matching ratio) and dividing by the sum of the denominators of Equations 2-5 (104). The test statistic used for evaluating the null hypothesis, A Hq J 0R*=1, was the Mantel-Haenszel chi-square, with 1 degree of freedom, and test-based confidence intervals were constructed where appropriate (see Appendix B). For the present study, all matched groups odds ratios, chi-square test statistics, and confidence intervals were computed by use of a program developed by OHSG personnel and based on formulae originally given by Mantel and Haenszel (104) and Miettinen (102). Both Mantel-Haenszel matched sample and crude estimates of the odds ratio for cases compared to controls were computed for each OTG. Three minimum duration of OTG-specific employment criteria were considered for defining exposure: 1. exposed - worked for 15 or more days in the OTG of interest; not exposed never worked in the OTG of interest or worked for fewer than 15 days. 2. exposed = worked for 2 or more years in the OTG of interest; not exposed *= never worked or worked for fewer than 2 years in the OTG of interest. 3. exposed *= worked for 5 or more years in the OTG of interest; not exposed = never worked or worked for fewer than 5 years in the OTG of interest. . GTR 000174 85 4 411 subjects must be classified according to exposure into dichotomous exposure categories in order to perform a matched analysis. For comparing "unexpoaed" subjects with those exposed for 2 or more years (or 5 or more years), two alternatives for classifying subjects as "unexposed" were considered: (1) subjects who worked in the OTG of interest for fewer than 2 years (or 5 years) were classified as not exposed and included in . subsequent analyses; (2) subjects who worked in the OTG of interest for some period of time that was less than 2 years (or 5 years) were excluded from consideration, and OTG-specific odds ratios were based on subjects who were either never employed (not exposed) or employed for 2 or more . (or 5 or more) years (exposed) in the OTG of interest. 1 For most analyses involving computation of odds ratios for more than one duration of employment category, both of these approaches were used. The effects of including in the not exposed categories workers having short-term exposures were evaluated by comparing results obtained by application of the two above exposure classification methods for the 2+ years and 5+ years minimum duration of employment categories. It was ' expected that exclusion of workers having relatively short-term exposures <| would reduce the possibility of obtaining diluted risk estimates. For example, in deriving risk estimates for OTG-specific exposures of 2+ years, classifying cases who were employed in the OTG of interest, but for fewer than two years, as not exposed would bias the odds ratio downward; classifying as not exposed controls having this same exposure history would bias the odds ratio upward. Alternatively, if the null hypothesis that duration of exposure is the same for cases and controls Is true, odds ratio estimates should not be biased In either direction: i.e OTG not tru thai more mort cas is p spec the more equa cate the OTG. expoi detei OTG \ 1 men with lung Crude folic GTR 000175 i.e. , cases and controls having shorter duration of employment in the OTG of interest than required by the criterion should be classified as not exposed with the same frequency. If the alternative hypothesis is true that length of service in a particular OTG was longer for cases than for controls, controls would tend to be classified as not exposed more frequently than will cases, and the association between lung cancer mortality and exposure would appear to be stronger. Although the design of the study minimizes differences between cases n/1 controls in duration of overall rubber industry employment, it is possible that considerable differences between the amount of time speilt by cases and controls in each OTG are present. For example, while the proportion of cases and controls employed in a' given OTG for 2 or more years may have been the same and the resulting crude odds ratio equal to 1, the cases falling in this minimum duration of employment category may have had a mean work duration in the OTG of 10 years, while the controls may have spent only 3 years, on the average, in the same OTG. Such differences may be biologically meaningful, since duration of exposure is often considered to be a correlate of dose, and should be detected. Therefore, the mean cumulative duration of employment in each OTG was calculated separately for cases and controls who spent at least 1 month in the OTG of interest, and differences in means were evaluated with the two-sample t-test. In order to take into account the latent period associated with lung cancer development, year of first employment in each OTG was considered Crude odds ratios were computed for each OTG, with exposure defined as follows: GTR 000176 1. exposed = employed for at least 1 month in the OTG of i n t e r es and first employed there 15 or more years prior to death; not exposed = employed in the OTG of interest 1 month or first employed there fewer than 5 years prior to death. 2. exposed = employed in the OTG of interest 2+ years and first employed there 15 or more years prior to death; not exposed employed in the OTG of interest 1 month or first employed there fewer than 5 years prior to death. 3. exposed = employed in the OTG of interest 5+ years and first employed there 15 or more years prior to death; not exposed = employed in the OTG of interest <1 month or first employed there fewer than 5 years prior to death. The three resulting sets of odds ratios are thus based on allowances i the exposure definitions for increasing lengths of employment in each OTG and for a 15 year latent period. . Also, for each OTG the mean year of first employment in the OTG computed for cases and controls who had worked in the OTG for 1 month, 2+ years, and 5+ years. These means were compared, and differences were evaluated by the two-sample t-test. GTR 000177 CHAPTER V RESULTS AND DISCUSSION A. Reaults of the matching procedure The one hundred thirty-one lung cancer deaths occurring among white male hourly rubber workers between January 1, 1964, and December 31, 1973, were matched with 500 controls on year of birth (+ 3 years) and year of entry into the rubber industry (+ 1 year). As indicated in Table 7, 10 cases and 52 controls were subsequently excluded from the study. Table 8 gives reasons for exclusions of initially selected study subjects. One case was eliminated when no suitable control could be found. Nine cases and 23 controls were excluded because work history information either was not available or was Inadequate. An additional 26 controls were eliminated as the result of the withdrawal of their respective cases, and 3 controls both had missing work history information and had their respective case excluded. Therefore, study results are based on 121 cases and 448 controls. The configurations of the 121 matched groups are summarized in Table 9. After exclusions, there were 94 quintuplets, 21. quadruplets, 4 triplets, and 2 pairs. B. Comparability of study subjects with respect' to control variables Certain characteristics of the case series and the control group are presented in Table 10. As a group, cases and controls tended to have been born around 1900-1901, to have started working at the plant under study in 1930, and to have been about 29 years of age when initially employed. It is evident that the matching procedure was effective in GTR 000178 minimizing differences between cases and controls for these variables. Although the variations observed are slight, some of them are statistically significant, probably because of the large sample size involved. It should be noted that consideration of mean year of first hire - 1931 for cases and 1930 for controls - is somewhat misleading in terms of describing, this aspect of employment history. Examination of the frequency distributions of year of first hire given in Table 11 reveals that large proportions of both cases and controls entered the industry before 1930 and after 1940, while relatively few study subjects were hired during the 1930rs. This hiring pattern was probably the result of reductions of the rubber industry production activities that occurred .> . . ..A during the economic depression of the 1930rs. Inspection of the free distributions presented in this table further confirms that cases and controls entered the industry during similar epochs. As Indicated in Table 10 the mean difference between cases and _ their respective controls in duration of overall employment is not significant (p>0.3). Furthermore, examination of the frequency dis' of duration of overall employment for cases and controls given in Ta 12 also fails to reveal any substantial differences in length of c*_ . ' * 'si employment. 85.1Z of the cases, compared to 85.5 Z of the controls S' employed for 20 or more years. The fact that cases and controls w on the average, for the same amount of time suggests that cases controls had equivalent opportunity for OTG-specific employment, importance of this comparability with respect to the validity history-related analyses carried out in this study was discuss detail in Chapter IV. As previously mentioned, all subjects included in the present study were members of the 1964 Cohort of rubber workers. Most former employees, included in this cohort, worked for more than 10 years. Therefore, it was expected that the percentage of short-term workers (those having fewer than 10 years of employment) among both cases and controls would be small. Furthermore, cases and controls were matched on year of birth and year of first hire at the plant under study, these matching criteria resulted in corresponding similarity, between cases and controls in age at first hire. Thus, if most workers continued their employment at the plant until retirement at age 65 and if no strong respiratory carcinogen were present in the work environment, causing exposed workers to die of lung cancer before retirement, cases and controls should have had the same duration of employment. . Two possible alternative implications of the observed similarity between cases and controls in length of overall employment must be considered: (1) No potent lung carcinogen was present in the work environment. Therefore, both cases and controls tended to work until retirement age. Since the matching scheme used in this study insured that they were approximately the same ages at first hire, cases and respective controls had the same duration of employment. (2) A potent carcinogen was present, causing exposed workers to die of lung cancer soon after initiation of exposure. Then, because workers employed for fewer than 10 years may not have been included in the present investigation, the elevated risk for these short-term workers could not be detected. In order for this alternative to be plausible, one must hypothesize, that a carcinogen, potent GTR 000180 enough to cause rapid development of lung cancer and concomitant termination of employment within 10 years of intial exposure, was present. The existence of such a strong carcinogen seems unlikely. Furthermore, reported latent periods for lung cancer resulting from exposures to industrial carcinogens are usually greater than 15 years (see Table 1, page 19). ^ It is clear that, for workers included in the present study, there, is no association between duration of cumulative rubber industry employing and risk for lung cancer mortality. However, it should be emphasized ^ that this study was not intended to evaluate, the possibility of such an4 association, and, as noted above, efforts were made in designing the study to insure comparability between cases and controls in length of ; employment for the purpose of achieving valid estimates of OTG-specificP risk for lung cancer. , ... C. Place of Birth Place of birth distributions for cases and controls are given in?! Table 13. 11.5% of the cases were foreign-bom, compared to 11.4% of the controls. There is, therefore, essentially no difference in prdf of cases and controls who were b o m in foreign countries. This fact suggests that there is no association between lung cancer mortality^ foreign immigrant status in the population of rubber workers studie Table 14 gives place of birth distributions for cases and coni; b o m in the United States. A higher proportion of cases than of cc (34.9%, compared to 23.6%) was b o m in southern states, while a hi proportion of controls (18.3%, compared to 13.2% for cases) was bl Minimatlu i GTR 000181 It would be of interest in subsequent studies of rubber industry- workers to examine in detail the distribution of southern--born workers: according to work area, in order to determine whether the second mechan mentioned above may have had an impact on patterns of OTG-specific 1 cancer risk. The possibility that southern, migrant status may be a confounder of observed associations between work history and lung-cancer^ risk should also be considered. . D. Results of OTG-specific analyses Table 17 gives the number and percentages of lung cancer cases and controls employed in each of the 20 OTG's for at least one month. '*? . 4, Higher proportions of cases than of controls were employed in the folX OTG's: Milling (2), Extrusion (3), Calendering (4), Stock Preparation Reclaim Operation (14), Chemicals (15), Pliofilm Manufacture (16), and Special Products Manufacture (19). The proportions of cases employed in Compounding and Mixing (1) and in Curing (9) are slightly less than J the corresponding proportions of controls. With regard to the remaining OTG's, the proportions of cases and controls are either equal, or the" -3 proportion of controls exceeds that of cases. The largest differences-^ are observed for Milling (2), Extrusion (3), and Special'Products Manufacture (19), all of which showed a higher proportion of cases than'; or of controls employed for a least one month. For Extrusion (3), the " '^ Mdb difference is statistically significant at the 0.05 probability level. ^ .. ca None of the remaining observed differences approaches statistical em significance. co No case or control was employed in OTG 20. Therefore, no further 3 reference to this OTG will be made. GTR 000183 94 The numbers and percentages of lung cancer cases and controls employed in each OTG for a minimum of two years are given in Table 18. As in the preceding table, the proportions employed in Compounding and Mixing (1) and in Curing (9) are essentially equal for cases and controls. The elevation in proportion of cases employed in Extrusion (3) observed in Table 17 is not noted for those who worked for at least two years. In addition, the small positive differences between proportion of cases and controls employed in Milling (2), Calendering (4), and Stock Preparation (5) are either reduced or absent for those employed for two or more years. This indicates that cases who contributed to the elevated proportions seen in Table 17 were employed in these OTG's for fewer than two years. The previously noted elevated proportions of cases employed in Reclaim Operation (14) and in Special Products Manufacture (19) are again observed in Table 18. For Chemicals (15) and Pliofilm Manufacture (16), the proportion of cases employed remains higher than that of controls. The greatest positive differences are observed for Reclaim Operation (14) and Special Products Manufacture (19). However, none of the differences in proportions reported in this table is statistically significant. The number and percentages of cases and controls employed for five or more years are given for each OTG in Table 19. Compounding and Mixing (1) and Curing (9) show only slight differences in proportions of cases and controls employed. For Extrusion (3), the proportion of cases employed for five or more years is again greater than the proportion of controls. However, the numbers of cases and controls who worked in OTG 3 for at least five years are small, and the observed difference in GTR 000184 proportions is not statistically significant. The proportion of cases (9.1%) employed in Reclaim Operation (14) is more than twice that of controls (4.2%). This difference is statistically significant (p=0.04). Consistent elevations in proportions of cases, compared to controls, are M- observed for Chemicals (15), Pliofilm (16), and Special Products Manufactu^* (19). For these OTG's fewer subjects were employed for five or more years, and the magnitude of differences in proportions is reduced, relative to those observed in Tables 17 and 18. Relative risk estimates (adds ratios) for each OTG and for three minimum duration of OTG-specific employment are given in Table 20, The;'- odds ratios may be considered "crude", in the sense that they are derived-; without consideration of the matching procedure. For the purpose of computing these risk estimates, workers who fulfilled the stipulated ' minimum duration of employment criterion were regarded as exposed, while those who worked in the OTG of interest for any amount of time less than1 1 that given by the criterion were classified as not exposed. Thus, the comparisons on which the odds ratios given in this table are based are a analogous to those made in Tables 17, 18, and 19. The observed results-^ and tests of significance are also equivalent. ' Most of the odds ratios in Table 20 cluster around unity, indicate an absence of association between a history of employment in the OTG interest and risk for lung cancer. As expected on the basis of compari of proportions of exposed cases and controls and tests made for Tabl#,: 17, 18, and 19, the only odds ratios that are statistically signify elevated above one are those for Extrusion (3) for exposures of at one month and for Reclaim Operation (14) for exposures of 5+ years. 21 96 Also, for the latter OTG, the odds ratio increases from 1.3 (exposure > 1 month), to 1.7 (exposure>2 years), to 2.3 (exposure >5 years), suggesting a possible association between both occurrence and duration of employment in this OTG. For Compounding and Mixing (1) and Curing (9) , the odds ratios are uniformly close to one. The computation of the risk estimates reported in Table 21 differs from the procedure used for the odds ratios in Table 20 as follows: each odds ratio in Table 21 is based on comparison of those designated as exposed by the minimum duration of employment criterion with those never employed in the OTG of interest or employed cumulatively for less than one month. Comparison of those classified as exposed with workers never employed or employed in the OTG of interest for less than one month, rather than with subjects who may have worked for longer amounts of time, reduces the possibility of obtaining diluted risk estimates. Most of the odds ratios in Table 21 are close to one, again indicating no association between employment in the OTG of interest and risk for lung cancer. The odds ratio of 1.6 for those employed in Extrusion (3) for at least one month, compared to workers employed for less than one month, represents a statistically significant elevation in risk. -The odds ratio is also elevated for those employed in Extrusion for 5+ years, but this elevation is not significant. For Reclaim Operation (14), the odds ratio again increases from 1.3 for those employed > 1 month, to 1.7 for those employed 2+ years, to 2.2 for those employed for at least five years. The latter estimate is statistically significantly 2 elevated above unity (X -=4.24, p=0.04). Finally, consistently elevated risk estimates are observed for Chemicals (15), Pliofilm (16), and GTR 000186 97 Special Products Manufacture (19), although none of these achieved statistical significance. It does not appear that the risk estimates given in Table 20 for the 2+ and 5+ years minimum duration of employment classifications are biased towards the null by inclusion among the non exposed of workers having >1 month of employment. Mantel-Haenszel odds ratios for matched groups data are given in Table 22. In this table, risk estimates are based on exposed defined as employed in the OTG of interest for the amount of time stipulated by the minimum duration criterion, and not exposed defined as employed for any period of time Jess than that given by the criterion of interest. The odds ratio for Reclaim Operation (14), for those employed for at least r. 2 five years, is 2.2 (X *=4.52, p<0.05), and for Special.Products Manufacture' 2 . (19), for those ever employed, the odds ratio is 1.7 (X -4.77, p<0.05). . None of the other odds ratios included in this table is statistically significant at the 0.05 probability level. However, odds ratios having a 90% confidence interval with a lower limit that is greater than one are observed for the following OTG's: Extrusion (3), for exposures of * any duration and of 5+ years; Pliofilm (16), for exposures of any duration;' and Special Products Manufacture (19), for exposures of 2+ years. For ' Reclaim Operation (14), the tendency for the magnitude of the odds ratio to increase with increments in the minimum duration of employment criterion^ is again apparent. _ ' Because the definition of non-exposed for odds ratios given in Table 22 may include in this category subjects who worked in the OTG of interest, the observed risk estimates may be somewhat conservative as noted also for the analyses presented in Table 20. For this reason, Mantel-Haenszel odds ratios, with exposed defined as employed in a given OTG for two or more years or for five or more years, and not exposed GTR 000187 defi: the 1 are not cate the empl Exam esti rati woul odds com; Tab.' of < in ' of est tha are nec dur Hae mat Tal defined for each duration of employment category as never employed in the OTG, were derived. These are shown in Table 23. These estimates are based on smaller numbers of cases and controls, since all subjects not falling precisely into either the "exposed" or the "not exposed" categories were excluded from the analysis. For example, in computing the odds ratio for Extrusion (3), 5+ years of employment, all subjects employed in this OTG for one month through four years were excluded. Examination of the odds ratios in this table does not suggest that the estimates reported in Table 22 are biased towards the null. As explained in Chapter IV, it was expected that the crude odds ratios, which ignore the matching procedure employed in this study, would be biased towards the null with respect to the Mantel-Haenszel odds ratios, which take the matching into account. It is apparent from comparisons of results given in Tables 20 and 21 with those given in Tables 22 and 23 that this bias, in terms of both efficiency and magnitude of observed point estimates, is either slight or absent. For example, in Table 21 the crude odds ratio for Reclaim Operation (14), for exposure 2 of 5+ years, is 2.2 (X =4.24). In Table 23, the Mantel-Haenszel point 2 estimate is 2.1 (X =3.31). This lack of bias towards the null suggests that exposure (i.e., OTG-specific employment) and the matching variables are not highly correlated. However, because the matching procedure was necessary in order to insure equal opportunity for OTG-specific employment during comparable time periods for cases and controls, the MantelHaenszel odds ratios derived through stratified analysis based on the matched groups confirm the validity of the crude risk estimates given in Tables 20 and 21. GTR 000188 Table 24 gives crude odds ratios, with exposure defined by three minimum duration of employment criteria and an allowance for a 15-year latent period. These risk estimates differ from those computed for preceding tables as follows: (1) all subjects who worked for at least one month in a given OTG but who were first employed in the OTG 5 through 14 years prior to death were excluded from the analyses; (2) all subjects who worked for less than one month or who worked for up to five years in the OTG of interest but who were first employed there fewer than five years prior to death were considered as not exposed. Thus, exposed subjects for each minimum length of employment category .. must have started working in the OTG at least 15 years before year of death. Comparisons of odds ratios given in this table with those shown , in Table 21 indicate the following: (1) for Extrusion (3), for the>l month category, the odds ratio decreases from 1.6 (Table 21, no allowance for latent period) to -- 1.4 (Table 24, 15-year latent period stipulation). The observed decrease occurred because of the exclusion from .the latter analysi^j of seven cases who were first employed in Extrusion 5 to 14 years.; prior to death. Twenty-two controls were excluded for the same reason. In addition, two controls who first started working~in O' 3 fewer than five years prior to death were reclassified as not_ y exposed for the analysis presented in Table 24. Computation of/,,, risk estimate, basing the exposure definition on a latent periooj: allowance of five or more years, instead of 15 or more years, results in an odds ratio of 1.7 (X^=4.37, p=0.05). For the five; GTR 000189 100 year minimum duration of employment category, the odds ratio increases from 2.0 (Table 21) to 2.3 (Table 24). The observed increase occurred as the result of the elimination of six controls who had five or more years of experience in Extrusion but who first started working there 5-14 years before death. As expected, no cases or controls were reclassified as not exposed for this length of employment category. (2) for Reclaim Operation (14), for the >1 month category, the point estimate of the odds ratio decreases from 1.3 (Table 21) to 1.2 (Table 24) i This slight decrease occurred because of the exclusion of two cases, compared to one control, who started working in Reclaim (14) 5 through 14 years before death. In addition, one control who was first employed in OTG 14 within five years prior to death was reclassified as not exposed. For the 2+ years category, the odds ratio remains unchanged. For the 5+ years duration of employment category, the odds ratio increases from 2.2 (Table 21) 2 to 2.5 (X =5.37, p=0.02) (Table 24). One control, first employed ^ in Reclaim 5-9 years prior to death, was excluded. No case was eliminated or reclassified with respect to exposure status. (3) for Special Products Manufacture (19)' for the 2+ years minimum duration of employment category, the odds ratio increases from 1.7 (Table 21) to 1.8 (Table 24). Four controls and one case who . started working 5 through 14 years before death were excluded. In addition, four controls who worked for two or more years but who were first employed in OTG 19 fewer than five years prior to death were classified as not exposed. GTR 000190 loi (A) with regard to other OTG's, allowance for a 15-year latent period does not result in marked changes in risk estimates. In no Instance does the consideration of a 15-year latent period lead to the emergence of an association not already indicated by previous analyses conducted In this study. It was not possible to carry out separate analyses, based on odds 'try ratios, for mutually exclusive categories of latent periods shorter than 15 years, because of small numbers of subjects classified as exposed in such categories. In order to ascertain whether or not there were differences between cases and controls with respect to the years in which they were first employed in given OTG's the mean year of first employment in each OTG for those who worked for >1 month, 2+ years, and 5+ years were computed and are reported In Tables 25, 26, and 27. For Extrusion (3), cases started work within two years of controls, on the average. For Reclaim Operation (14), cases were first employed, on the average three years (for the >1 month and 2+ years categories) to five years (for the 5+ years category) before controls. The observed differences in mean year of first employment in OTG 14 are not statistically significant. For .J Curing (9), cases were first employed, on the average, four years (fot the >1 month duration of employment category) to seven years (for the !i and 5+ years categories) before controls. For those who worked for at ni least two years and for at least five years, the differences in meat!' " year of initial employment in OTG.9 Is statistically significant at 0.05 probability level. Finally, for OTG 15 (Chemicals), cases working, on the average 4-7 years later than controls. Apparently, cases tended to have been first employed In OTG 15 in the late 1940*|j early 1950's, while controls tended to have started working in In the mid 1940's. GTR 000191 102 The implications of the above observed differences in mean years of initial employment in Curing (9), Reclaim Operation (14), and Chemicals (15) cannot be objectively determined. However, the existence of such differences suggests that the exposure experiences of cases may have differed qualitatively and/or quantitatively from those of controls, since they were employed, at least in part, during different calendar periods. Table 28 gives mean durations of employment for each OTG for cases and controls having at least one month of experience in the OTG of interest. Although none of the differences in means observed is statistically significant, it 6hould be noted that cases' mean length of employment in Reclaim Operation (14) was nearly twice that of controls - 11.2 years for cases, compared to 5.7 years for controls. This information may be interpreted as further evidence for an association between a history of employment in this OTG and risk for lung cancer. Other OTG's in which cases spent a greater number of years than controls include Extrusion . ^ (3), Curing (9), Finishing, Inspection and Repair (10), Synthetic Latex Manufacture (17), and Metal Products (18) . E. Summary of results of the OTG-specific analyses The major findings of the OTG-specific analyses conducted in this study are summarized in Tables 29, 30, and 31. The Mantel--Haenszel odds ratios given in these tables are reported in Table 22. Compounding and Mixing (OTG 1) and Curing (OTG 9) The Mantel-Haenszel odds ratios for Compounding and Mixing (1) and for Curing (9) presented in Table 29 are unifor&ly close to unity and indicate that the exposure rates for these two work areas were the same rtr nnnib? >r lung cancer cases and controls and, hence, that, for the population of rubber workers studiedt there is no association between risk fpr lung cancer and a history of employment in either of these work areas. Thus, the current investigation fails to replicate findings of an association between increased risk, for lung cancer and employment in jobs involving the compounding and mixing and curing of rubber previously reported in the literature (4-10). Several plausible reasons for discrepancies across studies in findings regarding, the risk indicator status of the compounding and. mixing and curing vo_rk ateas may be advanced. First, potential expos urea in the work areas, designated, as compounding and mixing and curing may not have been comparable for all rubber worker populations considered in the various studies The compounding and mixing and curing work areas involve the potential for exposure to respiratory hazards at most rubber plants by virtue of the fact that known respiratory system irritants and-, perhaps, carcinogens were used in processes carried out in these areas. For compounding and mixing, respiratory hazards include pulmonary irritants such as zinc oxide, zinc stearate, carbon black, talc, and phenols and, possibly., carcinogens such as cadmium compounds (used as accelerators) and chromates present in pigments. For curing, potential exposures are to talc and phenol-formaldehyde resins, and to unknown reaction products present in curing fumes. However, there are no available environmental data indicating that the intensity of use of these materials was the same at all rubber plants studied; nor is there any information regarding possible differences in industrial hygiene practices and environmental control measures at the various plants. Variation in these factors could have influenced risk for lung cancer. . GTR 000193 Sea deaths, due to c Fifteen cancer. mixing for can Therefc contro.' Compou work a the cc shoul< disea eithe pres cons< repo indi dif oth occ no' 104 Second, controls for the present study were selected from among deaths, including all deaths due to other malignancies, and 76 deaths due to cancers other than lung cancer were accordingly chosen as controls. Fifteen of these were deaths due to prostate cancer and 4, to stomach cancer. Associations with a history of employment in compounding and mixing for cancer of the prostate and in OT's involving high talc exposure for cancer of the stomach have recently been reported (OHSG, unpublished). Therefore, inclusion of subjects who died of these malignancies in the control group for the present study may have biased effect measures for Compounding and Mixing towards the null. Future investigations of this work area, if conducted in the case-control mode, should exclude from the control group workers who died of these malignancies. Also, it should be noted that, if some other cause of death (cardiovascular disease, for example) were associated with a history of employment in either Compounding and Mixing or Curing in the control group used in the present study, the effect measures for these OTG's would, again, be conservative. . Third, the risk estimates for compounding and mixing and curing reported by previous studies may have been spuriously elevated, and the indicated associations may have been due either to chance or to the differential exposure of subjects employed in these two work areas to other risk factors, such as cigarette smoke. Several important non- occupational risk factors for lung cancer (potential confounders) were not controlled for in the present or in previously reported studies. GTR 000194 105 Reclaim Operation (OTG 14) Mantel-Haenszel odds ratios for Reclaim Operation (OTG 14) are summarized in Table 30. The observed risk estimates are consistently elevated above unity and increase with each increment in the minimum duration of employment exposure criterion. These odds ratios suggest that risk for lung cancer is associated both with the occurrence of employment in Reclaim Operation and with length of employment in this work area. The minimum duration of employment classifications do not, however, permit assessment of possible dose-response relationships, since exposure categories are not mutually exclusive. Among the 18 jtv cases and 54 controls employed in Reclaim Operation for at least 1 month (see Table 22), mean duration of employment in this OTG was nearly two ..f r * times longer for cases than for controls, and cases, on th average, ,,i;% were first employed in Reclaim Operation three years before controls. Thus, since duration of employment may be a correlate of dose and since changes in industrial hygiene practices may have resulted in improvements in exposure conditions over time, these latter two findings suggest that ' ,Jmt cases who worked in Reclaim Operation may have sustained heavier exposures to respiratory system hazards present in this OTG than controls. Respiratory; system hazards are numerous in Reclaim Operation and include potential heavy exposures to particulates released during the process of shredding scrap rubber products, to chemicals and oils added to the shredded rubber during devulcanizing, and to fumes from heated rubber emitted during milling. For exposures defined as 5 or more years of employment in Reclaim Operation, the 95% confidence interval (1.06, 4.57) of the odds ratio does not include one. Therefore, it is unlikely that the suggested GTR 000195 106 association between a history of employment in Reclaim Operation and elevated risk for lung'cancer is due to chance. It should be noted that McMichael et al. (10) also detected the presence of an association between employment in the Reclaim work area and death due to malignancies of the respiratory system. The relative risk reported by the latter investigators was 2.3 for workers having at least 5 years of exposure to Reclaim. This risk estimate suggests that the magnitude of the association observed for Reclaim was similar to that noted in the present study (OR = 2.2 for those employed 5+ years). Chemicals (OTG 15), Pliofilm (OTG 16), and Special Products Manufacture (OTG 19) Mantel-Haenszel odds ratios for Chemicals (15), Pliofilm (16), and Special Products Manufacture (19) are summarized in Table 31. For these OTG's, point estimates of relative risk were consistently elevated above one for each minimum duration of employment category. However, there is no evidence suggesting an association between duration of employment in these OTG's and risk. The mean duration of employment in each of the OTG's mentioned in this table is approximately equal for cases and . controls (see Table 28). Cases started working later than controls, indicating the possibility of differential exposures fqr cases and controls employed in these OTG's. With the exception of the risk estimate (5r = 1.7) for Special Products Manufacture (19), for those,ever exposed, the odds ratios for these OTG's are not statistically significantly - elevated. The numbers of workers employed in these OTG's are small, and the observed associations may have been due to chance. However, the possible excess in lung cancer risk for workers employed in Chemicals, Pliofilm, and Special Products Manufacture should b e 'investigated further, and attempts should be made to identify agents, if any, present' in these areas and having potential etiologic significance. GTR 000196 Tubes (OTG 6) and Product Fabrication (OTG 7) Finally, it should be noted that the odds ratios for Tubes (OTG 6) were 0.5 and 0.6 for those employed for 2+ and 5+ years, respectively. The Tube OTG includes jobs entailing potentially heavy exposures to talc. It has been suggested that workers exposed to industrial grades of talc may sustain an elevated risk for lung cancer mortality (66). However, the absence of an association between increased risk for lung cancer and a history of employment in Tubes in the present study does not support this hypothesis. Two explanations for this finding may be advanced: , (1) Workers with a history of employment in Tubes tended to: di$ of exposure-related causes other than lung cancer (e.g., chronic respiratory diseases). (2) Workers employed in Tubes tended to reduce their cigarette consumption, perhaps because of high levels of respiratory system irritants present in this work area. If talc does not contain respiratory carcinogens and is not a sufficient cause of lung .. cancer but, rather, contributes to the etiology of this disease .r. ;> only in the presence of carcinogens or cocarcinogens (such as thog^j contained in cigarette smoke), then the risk for lung cancer wnuldlii have been low for workers in Tubes. It is not possible to evaluate' the correctness of either of these assumptions in the present ..pbcr study. . The Product Fabrication OTG (OTG 7) includes tire building jobP^adh; r v*. Odds ratios for this OTG, reported in Table 22, are 0.7 and 0.8,.pjfr those employed for 2+ and 5+ years, respectively. Thus, the findings^ 45 the present study are nt indicative of an elevated lung cancer riskf--* workers engaged in tire building. Such an association was r e p o r t ^ ^ Mancuso (7) but has not been noted by other investigators (9,1(X.8^^? JOT GTR 000197 108 Table 7 Results of the Matching Procedure for the Rubber Worker I.ung Cancer Case-Control Study Number of subjects originally selected Number of subjects excluded Total number included Total 631 62 569 Cases 131 10 121 Controls 500 52 448 Table 8 Selection of Rubber Worker Lung Cancer Case-Control Study Subjects: Reasons for Exclusions Cases: Controls: Reason for Exclusion__________ ' Number Excluded No acceptable match found 1 Missing or incomplete work history_______9________ Total number excluded 10 Respective case excluded 26 Missing or incomplete work history 23 Both of the above reasons________________3 Total number excluded 52 GTR 000198 Table 9 Results of the Matching Procedure for the Rubher Worker Lung Cancer Case-Control Study: Configurations of Matched Groups 109 ~ : * ' ,'S ~ '. : . - 1^ r iVmi-ilm. ^ Xw3 Group (Number of controls Configuration______ per case)______ Quintuplets (A) Quadruplets (3) Triplets (2) Pairs (1) No Match CO) Total Number of Croups Number.of Grp] Before Exclusions After Excl\i3jLl 121 94 da 2 n 4 4 v3|| Sc 2 2 2 r- 131 121 -m ' GTR 000199 Table 10 Selected Characteristics of Lung Cancer Cases and Controls Characteristic Age at D e a t h ' Duration of Overall Employment (yrs.) Year of Birth Year of 1st Hire Age at 1st Hire Cases Group M e a n ( S .D.) 67.3 (8.9) 29.4 (9.3) 1901 (8.8) 1931 (11.8) 29.1 (9.2) Controls Group Mean (S.D.) 69.8 (8.8) Mean Paired Dif f e r e n c e (S.E. (D)) -2.47 (0.15) 29 .9 (9.2) 1900 (8.6) 1930 (11.7) 29.7 (9.1) i o CO CO -0.19 0.91 0.09 (0.48) (0.10) (0.04) (0.11 *Matching variable. % < 0.05 11 20 -16.8* - 0.4 * k // - 7.8 GTR 000200 / Table 11 Frequency Distribution of Year of First Hire for Lung Cancer Cases and Controls Year of First Hire <1910 1910-19 1920-29 1930-39 1940-49 1950-54 All Cases Number ( Z L ___ 1 ( 0.8) 29 (24.0) 39 (32.2) 5 ( 4.1) 44 (36.4) 3 ( 2.5) Controls Number ( % ) 2 ( 0.4) 117 (26.1) 146 (32.6) 12 ( 2.7) 166 (37.1) 5 ( 1.1) 121 (100.0) 448 (100.0) GTR 000201 Table 12 Frequency Distribution of Duration of Rubber Industry Employment for Lung Cancer Cases and Controls Du r a t i o n of Employment (yrs) ^10 10-14 15-19 20-24 25-29 ^ 30 All ' ^20 Cases Numbe r ( % ) 0 (0 ) 6 (5.0) 12 (9.9) 32 (26.4) 13 (10.7) 58 (47.9) 121 (99.9) 103 (85.1) Controls Number ( % ) 4 (0.9) 12 (2.7) 49 (10.9) 90 (20.1) 72 (16.1) 221 (49.3) 448 (100.0) 383 (85.5) GTR 000202 Table 13 Place of Birth Distribution for Rubber Worker Lung Cancer Cases and Controls Place of Birth U.S.A. Foreign Unknown Totals Cases Number (Per cent) 106 ( 87.6) 14 ( 11.5) 1 ( 0.8) 121 (100.0) Controls Number (Per cent) 395 ( 88.2) 51 ( 11.4) 2 ( 0.4) 448 (100.0) Table 14 Place of Birth Distribution for Native-Born Rubber Worker Lung Cancer Cases and Controls 38*4 vta. ' U2Cl W -esc? i% > ^ ^5 ... m Place of Birth Cases Number (Per cent) Controls Number (Per cent) Ohio 29 ( 27.4) 121 . ( 30.7) Southern States^" 2 Northeastern States 37 ( 34.9) 14 ( 13.2) 93 ( 23.6) 72 ( 18,3) Other U.S.^ 26 ( 24.5) 108 ( 27.4) Totals 106 (100.0) 397 (100.0) "''Includes the following states: Ky., Tenn., Ga., Ala., La ., Ark. Va. , N.C., S.C., Fla., Md., Tex.., Okla. 2 Includes the following states: Pa., N.J., N.Y. , Conn. 3 Includes the following states: W. Va., Ind ., 111., Nebr. , Mich. RTR 000203 114 Tabic 15 Place? of B ir tli: Chi-Square Test of Association, So uth er n S t a t e s vs. A 1 J. Other P l a c e s of liir th P l a c e of Birth Southern States All Other Cases N (% ) 37 (30.6) 84 (69.4) Co itt roi s N (z ) 93 (20.8) 355 (79.2) Total 130 439 Total 121 (100.0) 448 (100.0) 569 X2 = 5 . 2 1 1 4 , p < 0.025 T a b l e 16 Place of Birth: Chi-Square Test of Association, Northeastern States vs. All Other Places of Birth Place of Birth Northeastern States All Other Places Cases N (% ) 14 (11.6) 107 (88.4) Controls N (% ) 72 (16.1) 376 (83.9) Total 86 483 To tal 121 (100.0) 448 (100.0) 569 2 X 1 d .f 1.5045, p > 0.20 6TR 000204 115 Appendix A-2 OT - OTG Dictionary Occupational Title Group Component OT's 1. Compounding and Mixing 2. Milling 3. Extrusion 001, 002, 003, 004, 005, 0 0 6 , 0 0 7 , O O s 'j OTG 060, 065, 078, 083 009, 010, Oil, 012, 013, 079, 084 I 019, 021, 023, 024, 025 4. Calendering 014, 015, 016 5. Stock Preparation 017, 018 6. Tubes 020, 022, 031, 032, 034, 042 7. Product Fabrication 026, 027, 028 8. Curing Preparation 9. Curing 029, 030 033, 035, 036, 037, 038, 080, 085 - i 10. Einishing, Inspection & Repair 040, 041, 043, 081, 086 11. Maintenance 039, 044, 045, 046, 047, 048, 049, 050, 051, 052 12. General Service 053, 054, 055, 056, 057, 059, 082, 088 3 13. Shipping & Receiving 058 . 14. Reclaim Operation 061, 062, 063, .064 15. Chemicals 16. Pliofilm Manufacture 066, 075, 067, 076, 068,- 069, 077 070, 072, 073,- 4 j 1 V A 17. Synthetic Latex Manufacture 089, 090, 091 t-1 00 Metal Products 19. Special Products Manufacture 20. Miscellaneous 092, 093, 094, 095, 099, 100 - _ -*5 071, 087, 097, 098 111, 112, 113, 114, 118, 119 - ''*3 `"jS S - 0 . 1 . Exit 101, 102, 103, 104, 105, 106, 107, i o 3 r,y33 109,. 110, 115, 116, 117 ^ A p p en di x A-2 is re pro du c ed here to facilitate i n te r p r t t ion o f;'j OTG codes. GTR 000205 116 Table 17 Number and Per Cent of Lung Cancer Cases and Cont rols Ever* Employed in each of 20 OTC's 0TG 1 2 3 . 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Cases Number (%) .41 (33.9) 48 (39.7) 30 (24.8) 21 (17.4) 27 (22.3) 23 (19.0) 44 (36.4) 20 (16.5) 29 (24.0) 44 (36.4) 27 (22.3) 54 (44.6) 22 (18.2) 18 (14.9) 25 (20.7) 7 ( 5.8) . 2 ( 1.7) 10 ( 8.3) 31 (25.6) 0 ( 0.0) Controls N u mb e r (%) 153 (34.2) 142 (31.7) Chi-square^ 0.003 3.722 ( p = .06) 76 (17.0) 66 (14.7) 3-852 ( p = .05) 0.324 92 (20.5) 0.091 91 (20.3) 0.036 163 (36.4) 0,010 82 (18.3) 0.101 115 (25.7) 0.070 170 (37.9) 0.045 125 (27.9) 1.247 201 (44.9) 0.003 " 81 (18.1) 0.012 54. '(12.1) - - 0.455 72 (16.1) 1.113 12 . ( 2.7) 1.967 12 ( 2.7) 0.100 36 ( 8.0) 0.011 84 (18.8) 2.788 (p = .09) 0 ( 0.0) - *Employed cumulatively f o r ^ x month. . #Chi -square with 1 d.f., for those e m pl oy ed, c o m p a r e d to those not employed in the OTC of interest. ^Note: no subject was employed in OTC 20. GTR 000206 Table 18 Number and Per Cent of bung Cancer Cases and Controls Employed for at Least 2 Years in each of 19 O T G 's " OTG 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Cases Num be r (%) 25 (20.7) 20 (16.5) 10 ( 8.3) 11 ( 9.1) 15 (12.4) 7 ( 5.8) 22 (18.2) 9 ( 7.4) 18 (14.9) 29 (24.0) 24 (19.8) 30 (24.8) 11 ( 9.1) 15 (12.4) 13 (10.7) 4 ( 3.3) 1 ( 0.8) 8 (6.6) 16 (13.2) Controls Number (%) ..Chi-- s a u; 87 (19.4) 0.031 77 (17.2) 0. 001 37 ( 8. 3) 0. 034 36 ( 8.0) 0.035 51 (11.4) 0.022 .54 (12.1) 3.284 110 (24.6) 1.828 41 ( 9.2) 0.168 70 (15.6) 0.004 88 (19.6) 0.842 97 (21.7) 0.095 112 (25.0) 0.005 48 (10.7) - 0.124 , ' 34 ( 7.6) - 2.800;| ' 30 ( 6.7) 1.692|I ' 8 ( 1.8) 0.45 7.1 4 ( 0.9) 0.23Q9 18 ( 4.0) 38 ( 8.5) 0.93 2 .49J Chi -s q u ar e wi t h ld.f., for those e m p l o y ed 2+ years., to those em pl o ye d for fewer than 2 years. . ; GTR 000207 118 Tn bIc 19 Number and Ter Cent of Lung Cancer Cases and Controls Employed for at L e a s t 5 Years In each of 19 OTG's OTG 1 2 3 4 . 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 C as es _____ N u m b e r (%) 17 (14.0) 15 (12.4) 8 ( 6.6) 6 ( 5.0) 12 ( 9.9) 5 ( 4.1) 18 (14.9) 5 ( 4.1) 15 (12.4) 20 (16.5) 22 (18.2) 23 (19.0) 9 ( 7.4) 11 ( 9.1) 7 ( 5.8) 3 ( 2.5) 1 ( 0.8) 5 ( 4.1) 4 ( 3.3) Controls Numb e r (%) 67 (15.0) Chi-square* 0.011 57 (12.8) 0.003 16 ( 3.6) 1.492 21 ( 4.7) 0.014 34 ( 7.6) 0.417 33 ( 7.4) 1.122 84; (18.8) 0.726 16 ( 3.6) 0.000 48 (10.7) 0.130 55 (12.3) 1.157 97 (21.7) 0.500 ^ 78 (17.4) 0.075 3.6 ( 8.1-) 0.001 19 ( 4.2) 4.487 (p=.04 ) 21 ( 4.7) 0. 067 6 ( 1.3) 0.232 3 ( 0.7) 0.185 13 ( 2.9) 0.155 10 ( 2.2) 0. 120 *Chi-square with 1 d.f., for- those employed 5+ years, compared to those employed for fewer than 5 years.. GTR 000208 119 Table 20 Odds Ratios for E m p l o y m e n t in 19 OTG 's by Three Minimum D u r a t i o n of Employment Categories Odds Ratios M i n i m u m D u r a t i o n of E m p l o y m e n t * .. OTG Ever 2+ years 5+ years 1 1.0 1.1 0.9 2 1.4 1.0 1.0 , 3 1.6 1.0 4 1.2 1 .1 1 *9 1 .1 e iVc 5 1.1 1.1 1.3 ;> M 6 0.9 0.4 . 0.5 7 1.0 0.7 0.8 8 0.9 0.8 1.2 9 0. 9 0.9 1.2 10 0. 9 1.3 1.4 11 0.7 0. 9 0.8 12 1.0 1.0 r. 1 13 1.0 0.8 . ` 0.9 14 1.3 1.7 - 2.3i 15 1.4 1.7 . I*2 16 2.2 1.9 1.9 - 17 0.6 0. 9 1.2 -H 18 1.0 1.7 1.4 . 19 1.5 1.6 1.5 *Any subject never employed in the OTG of interest or employed for a shorter period than that indicated by the minimum dura tion of employment criterion was considered "not exposed." Ifp -< 0.0 5 ` GTR 000209 \*A r . Sy! 120 Table 21 Odds Ratios for Those Employed, Co mpa re d to Those Never* E m p l o ye d , for 19 O T C 's by Minimum Duration of Employment Categories Three OTG 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Ifin imum Ever Odds Ratios Duration of Employment 2+ Years 5+ years 1.0 1.1 0.9 1.4 1.1 1.1 1.6 (X2= 3 .9) C 1.1 2.0 1.2 1.2 1.1 1.1 1.1 1 .3 0.9 0.5 0.6 1.0 0.7 0.8 0.9 0.8 1.1 0.9 0.9 1.1 0.9 1.2 1.3 "0.7 ' 0.9 0.8 1.0 1.0 1.1 1.0 0.8 . ' 0.9 . 1.3 1.7 - 2.2 ( X2= 4 .2) 1.4 1.7 1.3 2.2 1.9 1.9 0.6 0.9 1.2 1.0 1.6 1.4 1.5 1.7 1.6 *Never. for each du ra t io n of e m p l o y m e n t category,' is define d as zero or less than 1 month in the OTG of interest. ^Ever is defined as 1 or more m o n t h s of e m p l o y m e n t in the OTG of interest. cp <0.05. GTR 000210 121 T .t !>1 < 2 2 Odds Hat ios for M.'itehod Group:: for 19 OTG 1s by Titre Mini muni Durt: ion of K m p l o y m e n t Ca tep.o ries OTG 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 X8 19 Minimum Ever Odds Ratios Duration of Kmploymen 2+ years 5+ years 1.0 1. 3 // 1.6 1.1 1.0 1.1 0.9 1.0 // 2.2 m . ; 1.2 1.2 1.1 1.0 1.1 1.4 1.0 0 . 5 lf 0. 6 1.1 0.7 0.8 0.9 0.8 . 1.3 1.0 1.0 1.2 0.9 1.4 1.5 0.7 0.9 0.8 - 1.0 1.0 1.3 1.3 2.5* 0.7 0.9 1.7** 1.0 0.8 1.6 1.7 2.0 1.0 1.5 1.7* 1.1 i r.o 0.9 - ,,ftc 2 .2 " 1.3 2.0 1.3 1.5 1.5 *Any subject never employed in the OTG of interest or employed^ for a sh ort er period than that indicated by the m i n i m u m dura-:.* ,,tion of employment criterion was considered "not exposed. V 9 0 % c o n f i d e n c e limits do not i n c l u d e 1. . ^95% c o n f i d e n c e limits do not include 1. GTR 000211 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 125 Table 26 Mean Year* of First E m p l oy m e n t in 19 OTG's for Cases and Controls Employed for at Least 2 Ye ars in the OTG of Interest Mean Year of First Employment Cases (S .D .) Co nt rol s (S.D.) 1936 1936 O0.4) ( 9.8) 1939 1938 (11.1) (10.9) 1942 ( 9.5) .1940 (15.7) 1938 (12.7) 1937 (13.2) 1935 0 3 . 4 ) 1932 (12.1) 1928 ( 9.4) 1935 (11.2) 1932 (10.9) 1931 (12.4) 1941 (13.4) 1942 (13.7) 1926 (12.0) 1933 (12.3) 1931 (15.1) 1934 (14.7) 1939 (10.6) 1937 (12.5) 1936 (13.6) 1937 (13.3) 1936 (10.4) 1940 (11.4) 1933 (12.7) 1936 ( 1 2 . 2> 1949 ( 6.2) 1945 ( 8.0) 1954 ( 6.2) 1948 ( 7.4) 1946 (14.8) 1943 ( 5.7) 1943 ( 6.6) 1944 ( 5.1) 1943 ( 8.5) -- t -- -0.01 -0.64 0.51) 0.22 0.53 M . ;j? ::>3y . A; -1.83 0.23 .*&! * U- [' W `; - 2 . 3 4 (p < . o S) -1.05 0.98 i o M -"-J 0.06 0.36 ne are st year, PTC 1 2 3 4 5 6 7 8 9 10 11 12 13 14. 15 U i: 1! 1 * GTR 000215 126 Table 27 Mean Year* of First E m pl o y m e n t in 19 O T G 's for Cases and Controls Employed for at Least 5 Years in the OTG of Interest OTG 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Mean Year of First Employment Cases (S .I).) C o ntr ol s ( S .D . ) 1937 (10.4) 1935 (10.9) t 0.78 1935 (10.6) 1936 (11.1) -0.15 1939 ( 8.7) 1941 (16.8) -0.23 1935 (13.1) 1935 (14.7) 0.06 1935 (14.0) 1932 (12.2) 0.57 1926 (10.5) 1931 (10.0) -0.94 1930 (10.2) 1931 (12.5) -0.39 1943 (12.9) 1935 (11.2) 1.22 1923 (11.2) 1930 (12.5) - 2 . 0 7 ' (p<.05) 1927 (12.8) 1933 (15.5) -1.95 (p<.10) 1940 (10.3) 1937 (12.5) 1.23 1939 (12.8) 1937 (13.1) 0.61 " 1936 (11.4) 1938 (10.1) -0.45 . < 1932 (13.7) 1937 '(12.9) --1.12 1951 ( 5.7) 194.4 ( 8.2) 2.40 (p<. 05) 1956 ( 6.2) 1945. ( 5.2) 2.63 - - 1953 ( 3.5) - , ' 1943 ( 3.1) 1943 ( 5.7) 0.03 1947 ( 6.7) 1941 (10.2) 1.36 *To the nearest year. GTR 000216 127 Table 28 Me a n D u r a t i o n of Employment in 19 O T G' s for Lung Cancer Cases and Controls, Based on the Experience of Workers Having at Least 1 Month of E m plo ym ent in the O T G 's OTG 1 2 3 4 5 6 7 8 9 10 11 12 13 1A 15 16 17 18 19 Mean Duration of Employment (yrs.) t* Cases Controls 5.8 7.3 -1.59 5.A 6. A -0.75 A.8 3.7 0.71 6.2 ' 6.5 -0.13 7. A . 8.0 -0.28 00 o i A.8 6.6 7.1 9.9 -1.66 3.3 3.9 -0.59 9.7 7.8 0.87 7.5 5.6 1.35 17.1 18.2 -0. A A 6.2 7.0 -0.67 6.7 11.2 A.A 7.6 5.7 a ;6 - -0.A2 ' i.8 3 -0.1A as o i A. 7 7.0 10.7 3.1 6.3 3.5 0.50 6.8 1.05 2.6 .0.54 o o 'j ; ; .] r M 'll *Based on the co m p a r i s o n of group means; separate v a r i a n c e jaM estimates. GTR 000217 WSa 127 128 Table 29 Summary of M a n t e 1 - H a e n s z e 1 Odds Ratios for Compounding and Mixing and Curing I OTG iDura tio n of iEmploymen t Compounding & t M i x i n g (OTG 1) Ever 12+ ye a rs [5+ y ea rs N of Cases Exposed 57 25 17 Cur in g (OTG 9) [Ever 29 2+ years 18 15+ ye a rs 15 N of Controls ^ Exposed OR 2 90 7. X Cl 159 87 67 / 1.0 1.1 0.9 0.045 (0.67,1.37) 0.110 (0.70,1.73) 0.049 (0.55,1.58) 129 1.0 0.033 (0.66,1.40) 70 1.0 0.015 (0.58,1.62) 48 1.2 0.312 (0.71,2.05) GTR 000218 129 Tabic 30 Summary of M a n t e l - H a e n s z c 1 Odds Ratios for R e c l a i m O p e r a t i o n (OTG 14) M i n i m u m Duration N of Cases N of Cont rols 90% of Employ men t Employed Employed OR X2 Cl Ever 21 58 1. 3 1.139 ( 0 . 8 5 , 2 . 1 0 ) r 2+ years 5+ years 15 34 1.6 2.18 3 (0.95,2.66)* -o Z ' 11 19 2.2 4.522 (1.20,4.06.) * jChem: *95% confiden ce i n t er va l also does not include 1. |Ev er [2+ y 15+ y ' D Pli Eve GTR 000219 9 130 Table- 31 Summary of M a n t e l - H a e n s z e l Odds Rat i o s for Chemicals (OTG 15), Pliofilm (OTG 16), and Special Products M a n u f a c t u r e (OTG 19) .10)1 __3 .66)*3: .060 ^ ---- M 1 OTG Duration of Employment Chemicals (15) fever N of Cases Employed > 25 E + years 13 15+ y e a r s 7 IM svj p l i o f i l m (16) fever 8 |2+ y ears 4 [5+ y e a r s 3 N of Controls Employed 74 30 21 ' 13 8 6 90% OR X2 Cl 1.3 1.7 1.3 0.997 2.291 0.284 (0.83,2.08) (0.96,3.30) (0.59,2.80) 2.5 3.240 2.0 1.170 2.0 0.789 (1.08,5.94) (0.70,5.74) (0.55,7.22) Special Products M a n u f a c t u r e (19) fever 31 2+ years 16 p+ years 4 84 1.7 4.766 (1.14,2.61)* 38 1.7 2.886 (1.02,2,78) 10 -1.5 0 . 5 4 6 ( 0.60,3.89) i 95% confidence interval also does not include 1 . GTR 000220 CHAPTER VI CONCLUSIONS Epidemiologic studies conducted in Great Britain have demonstrated an excess in lung cancer mortality for rubber industry employees, compared to the general population of that country. In contrast, recent investigations of the mortality experience of rubber workers in the United States have not documented an excess in deaths attributed to lung cancer. A recent retrospective cohort study of 8,418 U.S. white male hourly rubber workers (83) reported a Standard Mortality Ratio of 80 for malignant neoplasms of the respiratory system for the years 1964 through 1973, indicating that the lung cancer mortality for these workers was lower than expected. The case-control study described in the present report was based on a lung cancer case series and control group selected from among deaths that occurred in this population of rubber workers and was carried out in order to determine whether history of employment in specific work areas in the rubber industry is associated with elevated risk for lung cancer mortality. The existence of such high risk work areas could not have been detected in the earlier retrospective cohort analysis, since detailed employment experience was not examined, and may have been obscured in a previous study (84) that focused on the work area in which subjects spent their longest amount of time and did not consider work area-specific employment of relatively shorter duration. Several previous epidemiologic studies that addressed the possibility that elevated risk for lung cancer is confined to certain work areas have documented increased risks for lung cancer among workers engaged in the compounding and mixing of raw rubber stocks and in the curing of GTR 000221 green rubber products. Associations between employment in these two work areas and high risk for lung cancer have been demonstrated in more than one study and, therefore, were of particular interest in the present investigation. However, it was also felt that the possibility of the presence of additional high risk work areas should be explored in the current study. The results of the present investigation do not support the findings of previous studies with respect to the compounding and mixing and curing work areas. Estimates of relative risk for these two work areas are uniformly close to one, indicating an absence of association between risk for lung cancer and history of employment in Compounding and Mixing and Curing, as defined in the present study. Cases employed for at least 1 month in Compounding and Mixing spent, on the average, a shorter period of time in this OTG than controls; among cases and controls employed for at least 1 month in Curing, cases worked only 2 years longer than did controls, on the average. Possible reasons for this apparent lack of association were advanced in Chapter V. It should be emphasized that the study design used for the present investigation may have resulted in conservative risk estimates. Thus', weak associations between lung cancer risk and employment in Compounding and Mixing and Curing may not have been detected. An association between elevated risk for lung cancer and history of employment in Reclaim Operation is suggested by the results of this study. The observed association is strongest for workers who spent at least five years in this work area and who started working there 15 or more years prior to death. Further analyses will be conducted in order GTR 000222 133 to determine more precisely the time period, if any, during which maximum elevation in risk occurred. Definition of this time period would facilitate the identification of possible etiologic agents and, if consistent with the concept of a 15-35 year latent period, would constitute additional supportive evidence for the presence of an association between employment in Reclaim Operation and elevated risk for lung cancer. The presence of comparatively high levels of particulates and fumes from heated rubber suggests that the observed association may be biologically plausible. Further work by industrial hygienists is necessary in order to fully characterize the nature and extent of possible etiologic agents present in the Reclaim Operation work area. The results of the present investigation suggest that there may be weak associations between history of employment in work areas designated as Chemicals (OTG 15), Pliofilm (OTG 16), and Special Products Manufacture (OTG 19). Little information is currently available with regard to agents, if any, that may have had an impact on the development of lung cancer among subjects who were employed in these work areas, and further work is needed to identify specific respiratory system hazards, if any, in these work areas. ' The validity of the findings of this study depends to a large extent on the correctness of the assumption that other lung cancer risk factors not controlled for by the matching procedure employed in the study design did not vary according to work area in the population of rubber workers studied. Risk factors other than OTG-specific employment, that were not considered include urban residence, previous employment in industries involving exposure to respiratory carcinogens, and, most GTR 000223 important, cigarette smoking habits. -The presence, extent, and direction of any bias due to lack of control for these potential confounders in the present study is unknown. It is not possible to refute the possibility that the positive associations detected in this study may have been the result of the differential, concomitant exposures of subjects employed in the putative high risk OTG's to nori-occupational respiratory system carcinogens, such as those present in cigarette smoke. Therefore, the results of the present study should not be regarded as conclusive. This study has not identified a strong association between history of employment in any one work area and substantial elevation in lung cancer risk. Furthermore, the facts that the mean ages at death for lung cancer cases and controls were nearly equal and that cases and controls tended to have worked for the same number of years suggest that no potent lung carcinogen wa6 present in the occupational environment of the population of rubber workers studied. It should be noted that the findings of the current investigation apply only to employees of the rubber plant under study. Because of the possibility of inter-plant variations with respect to the qualitative and quantitative nature of potential exposures in the various work areas considered and differences in industrial hygiene practices and environmental control measures, no statement regarding the work area-specific lung cancer risk among workers at other rubber plants can be made. Furthermore,' since short-term workers may not have been included in the present btfi - investigation, no inferences regarding their risk for lung cancer should , ' be drawn from results presented in this report. - * } . GTR 000224 135 The present study was conducted largely in the spirit of hypothesis generation rather than hypothesis testing. The purpose of the investigation was to identify high risk. OTG's at the rubber plant under study. For this reason, the tests of statistical significance reported in this paper may be inappropriate. Furthermore, a large number of such tests were conducted, and chance elevations in test results may have occurred. As previously noted, the OTG's considered in this study consisted of job groupings, based on similarities in process, product and location within the plant. No attempts were made to define specific jobs or exposures conferring the elevated lung cancer risks observed for certain OTG's. The nature and extent of environmental exposure hazards present in the jobs comprising these potentially high-risk OTG's require elucidation through further epidemiologic and environmental research efforts. GTR 000225 136 REFERENCES 1. S ilv e r b e r g C: C ancer S t a t i s t i c s , Ca 2 3 :2 - 2 7 , 1973. 2. Burbank F: United States Lung Cancer Death Rates Begin to Rise Proportionally More Rapidly for Females than for Males: A DoseResponse Effect? J. Chronic Disease 25:473, 1972. 3. Fraumeni JF: Respiratory Carcinogenesis: An Epidemiologic Appraisal. JNCI 55:1039, 1975. 4. Fox AJ, Lindars DC, and Owen R: A Survey of occupational cancer in the rubber and cablemaking industries: results of five-year analysis, 1967-71. Br. J. Industr. Me d . 5. Fox AJ and Collier PF: A survey of occupational cancer in the rubber ana cablemaking industries: analysis of deaths occurring in 1972-74. Br. J. Industr. Med. 33:249, 1976. ' 6 . Mancuso TF, Ciocco A, and El-Attar AA: An Epidemiological Approach to the Rubber Industry. JOM 10:213, 1968. 7. Mancuso TE: Epidemiological investigation of occupational cancers in the rubber industry. Paper presented to the International Conference on Occupational Cancers and Health Hazards in the Chemical and Rubber Industry, Switzerland, 1974. 8 . Monson RR and Nakano KK: Mortality Among Rubber Workers. I. White Male Union Employees in Akron, Ohio. Amer. J. Epid. 103:284, 1976. 9. McMichael AJ, Andjelkovic D, and Tyroler HA: Cancer Mortality among Rubber Workers: An Epidemiologic Study. In Occupational Carcino genesis, Ann. New York Acad. Sci., v. 271, 1976. - . 10. McMichael AJ, Spirtas R, Gamble JF, and Tousey PM: Mortality Among Rubber Workers: Relationship to Specific Jobs. JOM 18:178, 1976. 11. American Cancer Society, Cancer Facts and Figures, 1974, p.7. 12. Health Consequences of Smoking (1974) DHEW pub. #(CDC) 74-8704, p.47. 13. Burbank F: Patterns in Cancer Mortality in the United States: 1950-67. National Cancer Institute Monograph 33, U.S. Dept, of HEW, Public Health Service, pp. 199-216, 1971. 14. Murray JL and Axtell LM: Impact of Cancer: Years of Life Lost Due to Cancer Mortality. J N C I .52:3, 1974. - GTR 000226 137 15. Segi M: Cancer Mortality for Selected Sites in_24_ Countries. Tokyo Japan Cancer Society, NcT. 6 , 137, p. , 1966-67, 1972. ' 16. Schneiderman MA and Levin DI.: Trends in Lun Cancer, Cancer 30:1320 1972. ' ' ~* 17. Higgins HT: Trends .in Respiratory Cancer Mortality. Arch, Environ. Health 28:121, 1974. ' 18. Dol] R: Epidemiology of Cancer: Current Perspectives. Amer. J. Epid. 104:396, 1976. ~ 19. Mason TJ, McKay FW, Hoover R, et al_: Atlas of Cancer Mortality for U..S. Counties: 1950-69. DHEW Publ. No. (NIH) 75-780, Washington,~d7c . U.S. Gov't Printing Office, 1975. 20. Blot WJ and Fraumeni JF: Geographical Patterns of Lung Cancer: Industrial Correlations. Amer. J. Epid. 103:539, 1976. 21. Wynder EL, Covey LS, and Mabuchi K: Lung Cancer in Women: Present and Future Trends, JNCI 51:391, 1973. 22. Dorn. HF and Cutler SJ: Morbidity from Cancer in the United States. *" Public Health Monograph, No. 29:1-121, 1955. 23. Levin ML: The Occurrence of Lung Cancer in Man. ACTA International Union Against Cancer, V. IX, No. 29:1-121, 1953. 24. Third National Cancer Survey: incidence data. National Cancer Institute Monograph, No. 41. DHEW Pub. No.(NIH) 75-787. 25. W y n d e r EL: The Etiology of Lung Cancer. Cancer 30:1332, 1972 , 26. Fraumeni JF and Mason TJ: Cancer Mortality among Chinese Americans, 1950-67, JNCI 52:659, 1974. 27. Creagan ET and Fraumeni JF: Cancer Mortality among American Indians, 1950-67. JNCI 49:959, 1972 .' ` 28. Buell PE, Mendex WM, Dunn JE: Cancer of the lung among Mexican immigrant women in California. Cancer 22:186, 1968. 29. Stocks P: Recent epidemiological studies of lung mortality, cigarette smoking and air pollution, with discussion of a new hypothesis of causation. Brit. J. Cancer 20:595, 1966. 30. Haenszel W, Loveland DB, and Sirken MG: Lung cancer mortality as related to residence and smoking histories. I. White Males. JNCI 28:947, 1962. 31. Hammond EC and Horn D: Smoking and death rates - report on 44 months of foliow-up of 187, 783 men. II. Death rates by cause. JAMA 166:1294, 1958. ' GTR 000227 138 32. B u e ll.P : R e la tiv e im pact o f sm oking and a i r p o llu t io n on lun g c a n g e r. A rch. E n viro n. H ea lth 15:291, 1967. 3 3 . G r i s w o l d MH, W i l d e r CS, C u t l e r S.J, P o l l a c k ES: C n n c c j_ in _ C o n n e c t c u t , 1935-1951. H a rtfo rd , Conn., S ta te D ept, o f H e a lth , 955. 3-1. Z im m e r EG a n d H a e n s z e l W: C a n c e r i n _Towa. P u b l i c H e a l t h S e r v i c e , P u b l. No. 466, W ashington, ) / C . , U .S . G o v 't. P r in t in g O f f ic e , 19S6. 35. Hammond EC: Smoking habits and air pollution in relation to lung cancer. In Environmental Factors in Respiratory Disease, cd, II. K. Lee. New York, Academic Press, 1972, p. 177-196^ 36. Kotin P: Role of migrant populations in studies of environmental effects. J.Chron. Pis., 23:293, 1970. 37. Haenszel W: Cancer mortality among the foreign-bom in the United States. JNCI 26:37, 1961. 38. Dean G: Lung cancer in South Africans and British Immigrants. Proc. Roy. Soc. Med. 57:984, 1964. 39. Reid OC, Cornfield J, Markush RE, et aL: Studies of disease among migrants and native populations in Great Britain, Norway, and the United States. Nat'l Cancer Institute Monograph, 19:321-346, 1966. 40. Mancuso TF and Coulter EJ: Cancer mortality among native white, foreign-bom white, and non-white residents of Ohio: cancer of the lung, larynx, bladder, and central nervous system. JNCI^ 20:79, 1958. 41. Mancuso TF and Sterlin.TD: Relation of Place of Birth and Migration in Cancer Mortality in the U.S. - A Study of Ohio Residents (1959-1967). J. Chron. Pis., 27:459, 1974. 42. U.S. Public Health Service. Surgeon General's Advisory Committee on Smoking and Health, (1964). Smoking and Health, Public Health Service Publication No. 1103. * ~ 43. Wynder EL and Hoffman D: Experimental tobacco carcinogenesis. Science, 162:862, 1968. ' 44. Health Consequences of Smoking (1975), U.S. DHEW, Public Health Service. DHEW Publication No. (CDC) 76-8704, p. 7. 45. Kellerman G, Shaw CR, and Lyyten-KeHerman M: Aryl Hydrocarbon Hydroxylase Inducibility and Bronchogenic Carcinoma. New Engl. J. M e d ., 289:934, 1973. 46. Guirgis HA, Lynch HT, Mate T, et al.: Aryl Hydrocarbom Hydroxylase Activity in Lymphocytes from Lung Cancer Patients and Normal Controls . Oncology, 33:105, 1976. GTR 000228 4 7. McLcmore TL, M a r tin Rli, Busbee OL, e t a l . : A r y l h y d ro c a rb o n h y d r o x y la s e a c t i v i t y in pulm onary m arcophages and i ym phocytcs from lu n g ca nce r and ro n c a n c e r p a tie n ts . C ancer Resea rc h , 37:1175, 1977. 4 8 . M u l v i h i l l , J J : M ost f a c t o r s i n lu n g tu m o r s : An e x a m p le o f e c o g e n e t i c s i n o n c o l o g y . JN:C I , 5 7 : 3 , 1 9 7 6 . 4 9 . T a y l o r PH: The r e l a t i o n s h i p o f m o r t a l i t y and d u r a t i o n o f e m p lo y m e n t a s r e fle c te d by a c o h o rt o f chrom ate w o rk e rs . Amer. J . Pub. H e a lth . 56:218, 1966. " ' 5 0 . E n t e r l i n e , PE: R e s p i r a t o r y C a n c e r among C h ro m a te W o rk e rs . .16:523, 1974. JOM, 51. Bidstrup PL and Case RA: Carcinoma of the lung in workmen in the bichromates-producing industry in Great Britain. Brit. J. Industr. Med., 13:260, 1956. 52. Langard S and Norseth T: A cohort study of bronchial carcinoma in workers producing chromate pigments. Brit. J. Industr. Med., 32:62, 1975. -*53. U.S. Dept, of HEW, Public Health Service, CDC, NIOSH: NIOSH criteria for a recommended standard: Occupation Exposure to Chromium (VT HEW Publication # (NIOSH) 76-129, 1975. 54. Doll R, Morgan LG, Speizer FE: Cancers of the lung and nasal sinuses in nickel workers. Brit. J. Cancer, 24:623, 1970. 55. Pedersen E, Hogetveit AC, and Andersen A: Cancer of respiratory organs among workers at a nickel refinery in Norway. Int. J. Cancer, 12:32, 1973. 56. Lemen RA, Lee JS, Wagoner JK, Blejr HP: Cancer mortality among cadmium production workers, in Occupational Carcinogenesis. Ann. New York Acad. Sci., v. 271:273, 1976. 57. Blejer.HP and Wanger W: Inorganic Arsenic - Ambient Level Approach to the Control of Occupational Cancerigenic Exposures.* in Occupational Carciogenesis. Ann. New York Acad. Sci., v. 271:179, 1976. . 58. Lee AM and Fraumeni. J F : Arsenic and respiratory cancer in man: An occupational study. JNCI, 42:1045, 1969. 59. Ott MG, Holder BB, and Gordon HL: Respiratory cancer and occupational exposure to arsenicals. Arch. Environ. Health, 29:250, 1974. 60. Milham S and Strong,T: Human arsenic exposure in relation to a copper smelter. Environ. Res., 7:7176, 1974. 61. Cooper WC: Cancer mortality patterns in the lead industry, in occupational. carcinogenesis. Ann,New York Acad. Sci., v. 271:250, 1976. rtr nnn?29 140 62. Saffiotti U, Montcsano K, St-1 Inkumur AR, ct al.: Respiratory tract carcinogenesis in hamster inducted by di fforent numbers of administrations of BP (, Ferric Oxide. Can. Res., 32:1073, 1972 \ 63. Boyd JT, Doll R, Gaulds JS, ct al.: Cancer of the lung in iron ore (haematite) miners. Brit. J. Industr. Med., 27:97, 1970. 64. Selikoff IF, Hammond EC, and Churg J: Asbestos, Smoking and Neoplasia JAMA, 204:104, 1968. . ' * 65. Nicholsoi> WJ: Asbestos-The TLV Approach, in 6 ccupational carcinogenesis Ann. New York Acad. Sci., v. 271:152-169, 1976. 66. Kleinfeld M, Messite J, Zaki MH: Mortality experience among talc workers. A follow-up study. JOM, 15:345, 1974. 67. Rubino GF, Sconsetti G, Piolatto G, Romano CA: Mortality study of talc miners and millers. JOM, 18:186, 1976. 68. Lundin FE, Wagoner JK, and Archer VE: Radon Daughter Exposure and Respiratory Cancer. Quantitative and Temporal Aspects. Joint Monograph No. 10 (NIOSH-NIEHS) . U.S. Dept, of HEW, 1971. 69. Archer VE, Wagoner JK, and Lundin FE: Uranium Mining and Cigarette Smoking Effects on Man. JOM, 15:204, 1973. . 70. Wada S, Miyanish M, Nishimoto Y, e~t a l .: Mustard gas and a cause of respiratory neoplasia in man. Lancet, 1:1161, 1968. 71. Kuschner M, Laskin S, Drew RT, et al.: Inhalation carcinogenicity of Alpha Halo Ethers. III. Lifetime 5 Limited Period Inhalation Studies with Bis (Chloromethyl) Ether at 0.1 ppm. Arch. Environ. Health, 30:73, 1975. 72. Figueroa WG, Raszkowski R, and Weiss W: Lung Cancer in Chloromethyl Methyl Ether Workers. The N. Eng. J. Med., 288:1096, 1973. 73. Weiss W and Figueroa WG: Characteristics of Lung Cancer Due to Chloromethyl Ethers. JOM, 18:623, 1976. - 74. Waxweiler RJ, Stringer W, Wagoner JK, et at.: Neoplastic Risk among Workers Exposed to Vinyl Chloride, in Occupational Carcinogenesis. Ann. New York Acad. Sci., v. 271, 1976. 75. Tabershaw IR, Gaffey WR: Mortality Study of Workers in the Manufacture of Vinyl Chloride and its Polymers. JOM, 16:509, 1974. 76. Lloyd, JW: Long-term mortality study of steelworkers. V. Respiratory cancer in coke plant and its Polymers. JOM,16:509, 1974. 77. Doll R, Fisher REW, Gammon EJ, et al.: Mortality of gas workers with special reference to cancers of the lung and bladder, chronic bronchitis, and pneumoconiosis. Brit. J. Industr. Med., 22:1, 1965. GTR 000230 141 7S. D o l l R , V e s s e y MR, B e a s l e y RW, c t a l . : M o r t a l i t y o f g a s workers- f in a l r e p o r t o f a p ro s p e c tiv e s tu d y .. B r i t . J . In d . M e d ., 29-39-j 1972. * ' ' 7 9 . K a u a i M, A m a in o to H, a n d l l a r a d a K: I : j i d e m io lo p . i c s t u d y o f o c c u p a t i o n a l l u n g c a n c e r . A r c h . F i n v i r o n . H e a l t h , 14 :8 59 -8 (> -1 , 1 9 6 7 . 8 0 . U . S . D e p t , o f HEW: M o r t a l i t y i n 1 9 5 0 b y O c c u p a t i o n a n d I n d u s t r y . Vital Statistics - Special Reports, S3:Nos 1-5:91, June 1961 - Sept. 1963. ~ 81. U.S. Dept, of HEW: Occupational Characteristics of Disabled Workers, by .Disabling Condition, Disability Insurance Benefit Awards Made in 1959-62 to Men under Age 65, U.S. Gov't. Printing Office, Public ` Health Service Bulletin, No. 1531, 1967. 82. McMichael AJ, Spirtas R, and Kupper LL: An Epidemiologic Study of Mortality Within a Cohort of Rubber Workers, 1964-72. JOM, 16:458, 1974. 83. Andjelkovic D, Taulbee J, and Symons M: Mortality Experience of a Cohort of Rubber Workers, 1964-73. JOM, 18:387, 1976. 84. Andjelkovic DA, Taulbee J, Symons M, Williams T: Mortality of Rubber Workers with References to Work Experience. J O M , in press. 85. McMichael, AJ: Standard Mortality Ratios and the "Healthy Worker Effect": Scratching Beneath the Surface. JOM 18:165, 1976. 86. Fox AJ and Collier, PF: Low mortality rates in industrial cohort studies, due to selection for work and survival in the industry. Brit. J. Prev. Soc. Med., 30:25, 1976. 87. Histological Typing of Lung Tumors, Geneva, Switzerland, WHO, 1967. 88. Kreyberg L: Histologic Types of Lung Cancer - a morphological and biological correlation. Acta. Pathol. Microbiol, Scand. (suppl.), 157:15-64, 1962. ' 89. Shinton NK: Differences in Biological Characteristics o f Various Histological Types of Lower Respiratory Tract Tumours. Brit. J. Cancer, 17:1, 1963. 90. Harris CC: The Epidemiology of Different Histologic Types of Brochogenic Carcinoma. Cancer Chemotherapy Reports, Part 3, 4:59, 1973. 91. Whitwell F: The Histopathology of Lung Cancer in Liverpool: A Survey of Bronchial Biopsy Histology. Brit. J. Cancer, 15:429, 1961. 92. Yesner R, Gerstl B, and Auerbach 0: Application of the World Health Organization Calssification of Lung Carcinoma to Biopsy Material. Ann. Thorac. Surg, 1:33, 1965. 93. 94. 95. 96. 97. 98. 99. 100. 101 102 103 10* 10 10 1C GTR 000231 14] 'itiotml y - i a is. Cancel urvev 1th 142 9 3 . Y c s n e r R, C c lfm a n NA, a n d F c i n s t e l n AR; A R e a p p r a i s a l o f H is to p a th o lo g y in Lung C ancer and C o r r e la t io n o f C e ll Types w ith A n te ce d e n t C ig a r e tte Sm oking. A u e r. Rev. Rcsp. D is ., 107:790, 1973. ' 94. Berg J : E p ide m io lo g y o f th e d if f e r e n t h is to lo g ic types o f lu n g ca n g e r, in M orphology o f Exp e rim e n ta l R e s p ira to ry C a rc in o g e n e s is . AEC S y m p o s i u m , S e r i e s 2 1 . e d s . , N e t t c s h e i m P , H a n n a M:, a n d D eatherage J . 1970, pp. 93-104. 95. Whitwell F: The Histopathology of Lung Cancer in Liverpool: The Specificity of the Histological Cell Types of Lung Canger. Brit. J. Cancer, 15:440, 1961. 96. Weiss W, Boucot KR, Seidman H, and Carnahan WJ: Risk of Lung Cancer According to Histologic Type and Cigarette Dosage. JAMA, 222:799, 1972. 97. Auerbach 0 , Garfinkel L, and Parks VR: Histologic Type of Lung Cancer in Relation to Smoking Habits, Year of Diagnosis and Sites of Metastases. Chest, 67:382, 1975. 98. Whitwell F, Newhouse ML, and Bennett DR: A study of the histological cell types of lung cancer in workers suffering from asbestos is in the United Kingdom. Brit. J. Industr. Med., 31:298, 1974. 99. Saccomanno G, Archer VE, Auerbach 0, Kuschner.M, Saunders RP, and . Klein MG: Histologic Types of Lung Cancer Among Uranium Miners. Cancer, 27:515, 1971. 100. Cornfield, J . : A method of estimating comparative rates from clinical data. Application to cancer of the lung, breast and cervix. JNCI, 11:1269, 1951. 101. Cornfield, J and Haenszel, W. : Some Aspects of Retrospective Studies. J. Chron. Pis., 11:523, 1960. 102. Miettinen, 0.: Estimability and Estimation in Case-Referent Studies. Amer. J. Epid., 103:226, .1976. 103. Gamble, J. and Spirtas, R. : Job Classification and Utilization of Complete Work Histories in Occupational Epidemiology. JOM, 18:399, 1976. 104. Mantel, N. and Haenszel, W. : Statistical Aspects of the Analysis of Data from Retrospective Studies of Desease. JNCI, 22:719, 1959 105. Seigel, D. and Greenhouse, S.: Validity in Estimating Relative Risk in Case-Control Studies. J. Chron. Pis., 26:219, 1973. 106. Miettinen, 0. : Estimation of Relative Risk From Individually Matched Series. Biometries, 26:75, 1970. 107. F le is s , J . L . : S t a t i s t i c a l M ethods f o r R ates and P ro p o rtio n s - J o h n W i l e y a nd S o n s , New Y o r k , 1973 108. A rm ita g e , and Sons, P . : S t a t i s t i c a l Methods i n New York, 1971. GTR 000232 M e d ica l R ese arch . John W ile y appendix a- i M a s t e r O c c u p a t-ioiiiil T i t l e L i s t 143 Kunube r N ul 001 Batch prepa ra tx an -Ti rer; 002 Batch Preparation-Tubes, Fl ap; : , Bl a d d e r 003 Pigpent Blending 00 4 Cement Mixing 005 Cutting & Milling-Tires 006 Cutting & Milling-Tubes 007 Service-Batch Preparation 008 Service-Batch Preparation 009 Killing-Treads 010 Killing-Plystock Oil Killing-Flaps & Bladders 012 Killing-Tubes . 013 . Milling-Miscellaneous 014 Calender Operation 015 Calender Tending 016 Roll Changing-incl. Tracking & Service 017 Plystock Handling-incl. Band Building 018 Liner Service (reroll, mend, clean) 019 Tuber Operation-Treads 020 Tuber Operation-Tubes - 021 Tuber Operation-Flaps Bladders ' 022 Tuber Service-Tubes -- 023 -Tuber Service-Flaps Bladders 024 Tuber Service--Tread Tuber (incl. booking slitting) 025 Cerr.enting-Treads 026 Bead Euilding 027 Tire Building 028 Service-Tire Bead Building 029 Inspection, Repair, & Sort Green Tires (incl_ trucking] 0 30 Paint & Line-Green Tires 031 Valve Preparation 032 Tube Flap Building - . 033 Curing-Tires 034 Curing-Tubes ~ 035 Curing-Flaps (black) ' 036 Curing-Flaps (white) 037 Curing-Bladders ' 038 Curing-Valves 039 Mold Cleaning & Repair 040 Finishing & Inspecting-Tires 041 Repairing Tires 042 043 . 044 Finishing & Repairing-Tubes & Airbags' Finishing & Repairing-Flaps, Bladders, Maintenance-Painting ' Sleeves .i is 045 Main tenance-Mi llv/righ t 046 Maintenance-Electrical 047 Maintenance-Sheet Metal 048 Maintenance-Welding 049 Mechanic 050 Carpent-ry - Z tt 0 0 0 0 o: o: o: o: o: 0i 0i Oi 0< 0i 0< 0! o. 0! O 0 O' O' O' O' O' O' O' O' 0 o: 0i . o: 0: 0 0 0 0 0 ' 'O' ' 0 0 'o 0 0 0 0 0 0 0 1 c i t o nnn?T5 143 ing} uckii Appc-ii:! i >. A--.1 , c c - t. . 144 Number Machinist Maintenance-Misc. (pipefitters, cement f:i.ni5;her, e Lc:_) Power Plant Quality Control Testing Trucking (genl.) Clerical' Janitor Shipping Receiving Other Rubber Worker (tires tubes) Mill Mixing ' 061 P r e p .-Reclaim Devulcanizing-Reclaim Milling-Reclaim (refining) Service-Reclaim Service-BB-Misc. Cpd. Mix-Chem. Prod. Reaction Opr.-Chem. Prod. Finishing (dry, bag) -Cheiu. Prod. Ship. Rec.-Chem. Prod. . Manufacture Foam Prod. ' Rocket Liner Mfg. . Urethane Vinyl Prod. Mfg. Resin Prod. Mfg. Pliofilm Mixing (chemigurri, pliolite, - dispersite) 345A split Pliofilm Fabrication Pliofilm Finishing Batch' Prep.-Ind. Prod. Prep.-Ind. Prod. Curing-Ind. Prod. Finish Inspect-Ind. Prod. Other Rubber Worker-(non-tire tube) Batch Prep.-Hose Belt " Fabrication-Hose Belt . Curing-Hose 6 Belt . - Finish Inspect-Hose Belt Special Rubber Prod. Development (381A 357A) Other Worker (nori-rubbar-such as police, cafeteria ,'e t c . Batch Prep.--Synthetic Latex Preparation Reaction-Synthetic Latex Finishing Packing-Synthetic Latex Rim Plant-Metal Treatment - Rim Plant-Forming Machining Rira Plant-V7elding ' Rim Plant-Finishing Inspecting Plastics-Production Manufacture Fuel Cell , Aerospace Metal Preparation-(mostly degreasing operations) Mechanical Building Append ix A - l , c o n e . Number 101 102 10 3 104 105 106 10 7 108 109 110 111 112 113 114 115 116 117 118 119 Salciry Unknown Inaciive File Retired Sick Suspense Layoff ' Active Dead i.eave of Absence Kisc. Absence _ Anode Process (Coagulation dip) Rubber Preparation . ' Retread . Adhesives Exit From Work for Other Reasons Disability Retirement - Absence for Military Service Tube Building .& Curing ' .Footwear 145 i 1 1 ] RTR 000235 146 Appendix A-2 145 OT -- OTG Dictionary Jccupational Title Group 1- Compounding and Mixing [ 2. Milling 3. Extrusion 4. Calendering 5. Stock Preparation 6. Tubes 7. Product Fabrication 8. Curing Preparation 9. Curing [10. Finishing, Inspection & Repair |n. Maintenance [12. General Service I13. Shipping & Receiving 14. Reclaim Operation 15. Chemicals J f 16- Pliofilm Manufacture [17. Synthetic Latex Manufacture Metal Products ^ 19. Special Products Manufacture V 20. Miscellaneous 21. Exit Component OT's 001 , 002 , 003 , 004 , 005 , 006, 007,, 008, 060 , 065,, 078 , 083 009.. 010,, on,, 012., 013,, 079, 084 019,, 021,, 023,, 024,, 025 o H 015, 016 017, 018 020, 022, 031, 032, 034, 042 026, 027, 028 029, 030 033, 035, 036, 037, 038, 080, 085 040, 041, 043, 081, 086 039, 044, 045, 046, 047, 048, 049, 050, 051, 052 053, 054, 055, 056, 057, 059, 082, 088 058 - 061, 062, 063, 064 066, 067, 068, 069, 070, 072, 073, 074, 0 075, 076, 077 089, 090, 091 092, 093, 094, 095, 099, 100 071, 087, 097, 098 H I , 112; 113, 114, 118, 119 - 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 115, 116, 117 GTR 000236 147 Appendix A-3 SUnHAItY D ES CR IP T IO N 0_F OCCU P A T IO N A l. _TITLE_ CROUPS FOR THE L UNG CANCER CA S K - C O N T ROL STUDY 1. Com pounding & M ix in g Natural and synthetic rubbers, accelerators, antioxidants, fillers, etc. are weighed and then mixed in banburies for the preparation of various stocks in making tires, tubes, flaps and bladders- Service personnel truck bulk materials and blended pigments to banbury operators for dispensing into the banbury. Cements for use throughout the plant are mixed in'a separate area. Cements are solvents made of petroleum derivatives such as paraffin hydrocarbons (e.g. hexane, heptane, octane), aromatic hydrocarbons (e.g. toluene, xylene), chlorinated hydrocarbon (e.g. trichloroethylene), ketones (e.g. methyl ethyl ketone). Environ mental exposure to particulates (e.g. carbon black, accelerators, antio xidants, talc) is higher than other areas of the plant. Exposure to solvents is expected where cements are prepared. In the Cutting and Hil ling OT's (OT's 005 and 006), various batches from banburies are milled into long sheets, dipped in talc solution, dried and stored. Particulate exposures may be similar in regard to certain ingredients to those in other OT's included in this OTG, with the possible additional exposure to rubber fumes and reaction products from the hot uncured rubber stock. 2. Milling Rubber stocks are processed further under heat and pressure (on a mill) for softness, and a plastic state. Reaction products from the hot uncured rubber stock are the main contributors for environmental exposure. For O T 's 079 and 084, there are exposures to particulates (e.g. talc) and solvents. 3. Extrusion Rubber goes through a tuber (die) and is extruded as tread rubber. Tread rubber is then slit and cemented. Environmental exposures include reaction products and cement vapors. 4. Calendering The rubber from the mill is rolled through the calenders into sheets (usually covering a fabric). The calendar stock is then cut and spliced for further use. Also included in this group is roll changing (changing rolls and trucking rolls to storage, etc.). There is potential exposure to reaction products from the heated stock- 5. Stock Preparation . Splicing of material for making tire plystock and bands and rerolling GTR 000237 fa b r groti te n t 6. fabi and (in cur: 7. pro to 8. rep in age um am 10 an e? i: wi V. 1 j c ] Appendix A - 3, cent. 148 fabric (.liners) after the p.lyxtock is used is the pr.unary operation in chi.' group. Also cleaning nnd mending of J iners when necessary. There is po tential exposure to reaction products and some sol vent 6. Tubes This involves the extrusion of tube, flap, and bladder rubber, the fabrication of tubes, flaps, and valves, and the curing and finishing and repairing of tubes. Environmental exposures are to talc, solvents (in valve preparation, OT 031), reaction products and fumes j(in tube curing, OT 034) . 7. Product Fabrication (Tires & Beads) This involves all processes with building tires and beads and all processes in which tire and bead builders are serviced (trucking materials to builders). Exposures are to solvents and uncured rubber stock. 8 . Curing Preparation This process includes inspection of uncured tires for defects and repairing and spraying materials onto uncured tires to prevent sticking in the curing mold. Solvents and mold release agents are possible agents for environmental exposures. ' S Curing The uncured products (tires, flaps, bladders, and valves) are cured under heat and pressure. Environmental exposures are to reaction products and fumes. ' ' ' 10. Finishing & Inspection & Repair Trimming, balancing, labelling, buffing, repairing, force grinding, and classifying cured products are the major operations. Environmental exposures are mostly in buffing and repairing (solvents, and rubber dust). 11. Maintenance Mold cleaning, trucking, mechanics, pipefitting, electrical work, welding, painting, carpentry, etc., are jobs throughout the plant. Varied exposures as a particular job may indicate- 12. General Service . All operations associated with power plant, quality control, trucking, janitors, clerical, etc. This is a miscellaneous group with a variety of different jobs and exposures. 13. Shipping & Receiving All activities associated with receiving and unloading of raw materials and packing and loading finished products for shipping are included in this GTR 000238 Appendix A-3, cone. 149 category. An)' exposures in receiving would generally be. sporadic due to occasional leaks. There is little-, exposure in shipping. 14. Reclaim Operation Scrap vulcanized rubber products are shredded for separation and sizing before the rubber is "devuIcanized". Hy the application of heat and chemicals, the rubber compound is restored to its original plastic state by milling. Exposures are to particulates and fibers from shredding scrap rubber, chemicals and oils in devulcanizing areas and rubber fumes from milling. 15- Chemical Products Bulk chemicals are brought into the plant and stored in tank areas. Chemical products are produced in reaction tanks generally in a closed process. The chemical products are finally put through a drying procedure and bagged. They are now ready for plant use or for shipping. Exposures are to chemical solvents and chemical particulates. 16. " Pliofilm Manufacture _ Natural rubber is masticated and mixed with other compounds in a banbury. The stock is next transferred to mixing tanks where solvents /^ire added. The rubber mix is pumped to reactor vessels neutralizing tanks and filtered prior to spreading. The mixture is spread to desired thickness, dried and rolled as a finished product. Environmental exposures are mainly to solvents. 17. Synthetic Latex Manufacture This is essentially a chemical plant making elastomers that approxi mate one or more of the properties of natural rubber. SBR (Styrene-butadiene) is the most important synthetic rubber. Others include neoprene, nitrile, ethylene-propylenediene, etc. Exposures are to the g a s e s l i q u i d s and vapors that are ingredients for the particular synthetic rubber being made and the dried product. .- 1 8. Metal Products Raw steel is cleaned in a pickling tank prior, to being shaped (by bending) and welded. Finished metal products are lastly inspected, repaired when necessary and spray painted. Exposures are to welding fumes and acid and solvent vapors. 1 9. Special Products Manufacture These products are metal and rubber stock, to varying degrees, and -include rocket liner manufacture, special rubber product development, uel cell, and aerospace products. There are various exposures as a particular job may indicate. ; 20. I ; 1 1 jobs, s i 21. I 1 GTR 000239 Ap j>e nclLx A- 3 , c o n ! . 150 20. Miscellaneous T h is is a m isce lla n e o u s group w o r k e r s i n v o l v e d i n a v a r i c ty of jo b s , n o t o th e rw is e c la s s ifie d . E n viro n m e n ta l exposures v a ry . 21. Exit This is a group constructed for exits from plants for various reasons. 6TR 000240 Ap pend ix B Additional F o r m a lac Used in Ana]yses for the Rubber Worker Lung Cancer C a s e - C o n t r o l Study 1. Paired sample, t-test Hq : D = 0 17 *f/,, D = the mean d i f f e r e n c e , C"^-) , w h e r e d is the difference bet w e e n the v a l u e of a given variable for a case and the average value among his respective controls, and n is the n u m b e r of m a t c h e d g r oups. Where the sample variance of ference, s 2 , = 2td ^ - (" jd ) d ______ " n -1 the dif - Reference: 108 2. Two s a m p l e t-test V *1 - *2 ' mean of a given variable for cases; x _ = mean of a given variable for controls. n l+ n 2~^ X1 X 2 (n1- l ) s | + ( n 2- l ) s 2 -L-K-L. n^+n 2-2 N ni n2 2 2 . where s^ and s^ are the sample vari ances for cases and controls, respec tively, for a given variable, and n^ and n are sample sizes for cases and controls, respectively. Reference: 108 0 . Ref 5 . R< GTR 000241 152 Appendix B , cent. 3 . .C.hjLr:..S q U.2..HC__t e s t , o f -, X i-l d.f. 2 m 2 ( 0 - E . i-I a ss or i t ; o n f o r n i-.y_._2_ tn h 7p s . 1/h c r e 0 . = observed cell . i frequency, ' expected cell frequency, m numbe r of rows . Reference: 107 4. C h i - s q u a r e test of a s s o c i a t i o n for the crude o d d s r a t i o . 2 X 1. d.f. 2 X =2.71 0.10 N(ad - be) ni nomi rao X2 =3.84 0.05 (see F i g u r e 2, p a g e 84, for explanation of notation.) 2 X = 6.63 0.01 Reference: 104 5. C h i - s q u a r e test of a s s o c i a t i o n fo r the m a t c h e d s a m p l e odds ratio, for variable ratio matched data. where R is the m a t c h i n g ratio; m is the total n u m ber of exposed s u b j e c t s in a g i v e n group; the N = the n u m b e r of m a t c h e d groups in whi ch a total of m subjects is expo se d , and f p and are d e f i n e d as in F i g u r e 3, p a g e 8 6 . For example, if R=4, the N are defined as follows: N= 1 N2 = f + 10 f11 + f 01 f 02,, N 3 = f l2 + f 03 ii hti + N, f 4 13 04 Reference : 104 GTR 000242 153 A p p e n d i x B, cone. 6 . C o m p u t a t i o n o f a t e s t - b a r.od c o n f i d e n c e _i_u_t e r v a l f o r odds r a ti o e s t i m a t e s . C o n f i d e n c e i n t e r v a l ( C l ) - KXI' ( I n O lO O . t . ^ ) ' where ( I n OR) i s t h e n a t u r a l l o g o f t h e p o i n t e s t i m a t e o f the odds r a t i o ; Z is a percentage point of the normal d i s t r i b u t i o n ( Z = l .96 for a 95% Cl; Z=1.645 for a 90% Cl); and X i s t h e M a n t e l - H a e n s z e 1 chi. Reference : 102. GTR 000243