Document 997KwZ4ayRE1aZ2g88o1q95NR

A Retrospective Cohort Study of Mortality and Cancer Incidence Among Chemists Shelia K. Hoar, Sc.D., and Sidney Peli, Ph.D. This retrospective cohort study examines mortality and cancer incidence among 3,686 men and 75 women who were employed as chemists in 1959. During the period 1964 to 1977, the male chemists experienced lower overall mortality than other salaried employees of the chemical company (198 deaths observed, 241.0 expected, 5MR = 82). Large deficits are seen in lung cancer and arterioscle rotic heart disease deaths. The chemists appear to be at slightly higher risk for death from malignancies of the col on (12 observed, 6.7 expected, SMR = 178) and from cere brovascular disease (15 observed, 10.8 expected, SMR = 138). The low overall mortality resulted in a larger than ex pected proportion of deaths due to cancer. Fewer than ex pected cases were diagnosed of cancer of all sites com bined (61 observed, 86.5 expected, SIR = 71) and of the lung (8 observed, 20.0 expected, SIR = 40). The incidence rates of melanoma and of cancer of the prostate are slight ly higher than expected, relative to the Third National Cancer Survey and the experience of nonchemists, respec tively, Among female chemists, deaths due to all causes and suicide occurred more frequently than expected. Possible explanations for the lack of anticipated excess risks and for the observed deficits are presented. Keports from the United States, Sweden, and Great Britain suggest there is excess mortality from certain types of cancer among chemists.1'9 However, the designs of the studies reported have some limitations. First, the populations were not restricted to persons employed as chemists. The American and British studies of chemical society members may include persons who were not oc- From ihe Department of Epidemiology, Harvard School of Public Health, 677 Huntington Ave,, Boston. MA 02115 (Dr Hoar), and Epidemiology Section, Medical Divuion, E 1 du Pont de Nemours and Co, Inc., Wilmington, DE 19898 (Dr Pell) Address communications to Dr Hoar. Environmental Epidemiology Branch, National Cancer Institute, La'ndow 3C06. Bethesda, MD 20205. Supported by grants from E I. du Pont de Nemours and Co., Inc., the Exxon Corporation, and the United States National Cancer Institute (5 P01 06373). cupationally exposed to chemicals, for example, admini strators for chemical companies.'3 * Secondly, several of the studies were based on proportional mortality rates, not absolute mortality rates.1'378 Apparent excesses in proportional mortality from cancer may reflect deficits in other causes of death.10 Thirdly, the Swedish study which was based on absolute mortality and attempted to iden tify subjects who worked in a laboratory included too few deaths to be conclusive.8 E. 1. du Pont de Nemours and Company, Inc., employs many chemists and was concerned over these reports. Du Pont had data available to identify many persons em ployed as chemists and .to generate cancer incidence and mortality rates. This paper describes an absolute mortali ty and cancer incidence study of chemists employed by that company. . Materials and Methods The Du Pont Company maintains a computerize-d per sonnel file which includes records of all salaried persons who were active employees on or after January 1, 1964. Records of hourly wage roll workers are not included in the file. The file contains demographic informatio n and work histories retroactive to 1959. From this file, the authors identified 3,713 men and 75 women who were employed in 1959 as chemists, chemical engineers, or research associates, or in occupations with similar job titles. Because of the small number of nonwhites, the cohort was restricted to 3,686 male and 75 female white chemists. It is difficult to define the job title "chemist" a.nd to characterize occupational exposure. Most chemisis are exposed to numerous substances that change cons tantly in quantity and in quality. Different chemists rarel Y sus tain the same exposures. This is especially true for a cohort of Du Pont chemists. Du Pont is an interna tional company with many different processes, products, and locations, and chemists are employed throughout the company. A large number are employed in research h and development at the Experimental Station in Wilmington, Raprintad from Journal of Occupational Mtdlclna July, 1M1,Volum 23, No. 7 pp. 465-501 JOM 1681 485 DBZ02008 91 Table 1. - Du Pont Department Groups and Products. Dime Plastic resins, libers and products Examples ol Products or Exposures Plastic resins, filaments, sheetings, photo graphic papers, chemicals and equipment, textile fibers ' Basic chemicals and finishes Paints, pigments, chromates, finishes, te traethyl lead, Iluorinated compounds, dyes, solvents, polymers tor non-plastic uses, dry cleaning fluid, sodium products, electronic products, plant protection chemicals, teed supplements, pharmaceuticals, basic chem icals and Intermediates Petrochemicals Acrylonitrile, explosives, industrial diamonds Elastomers Neoprene, accelerators, antioxidants and miscellaneous chemicals lor natural and synthetic rubber, coated labrics Research, development and engineering Atomic energy Administration Low-level radiation Del., while others work In research and development at the plants. Some are involved with pilot plants or semi works operations for new products and processes, and others serve in engineering or quality control positions at the production plants. This variety is further complicated by rotation of some chemists from plant to plant or to the Experimental Station. The same job title and code number can represent dif ferent tasks at different company locations. For instance, "technical investigator" could be a laboratory research position at one location, but an administrative position at another. A chemical engineer could do laboratory work at one site, but design work on a computer at another. A few tasks are consistent across locations. For example, a "pat ent chemist" is always an office job, but the position may be held by-someone who was once a laboratory research chemist. Du Pont Is organized Into Industrial and administrative departments. Within a department, the types of processes and products have some similarity. Because most employees stay within the same department throughout their careers despite transfers to several locations, ex posures were characterized by usual and "ever" depart ment affiliation. A Du Pont industrial hygienist identified the seven major department groups that existed during the study years, combining those that had undergone name changes and reorganization (Table 1).n The authors also identified 19,262 male and 673 female white "nonchemists'' who were active salaried employees in 1959 and were listed in the computerized personnel file on januaryl, 1964. Persons ever employed by Du Pont as chemists were excluded from the nonchemist group. Extracts of the personnel records were created and missing demographic data were obtained from noncom puterized employee rosters stored by the Du Pont Epidemiology Section. Cancer Incidence. -- Cancer cases were ascertalned'by review of the Medical Division cancer registry. The registry consists of cases gleaned from the Accident and Health Insurance (AHI) claims and life insurance claims. The AHI claims are submitted for any disability of eight or more consecutive days. The diagnosis on the disability claim is certified by the attending physician. About 97% of employees with at least six months of service are enrolled in the AHI plan and all employees are enrolled in the life insurance plan. The registry and insurance plan have been described in detail in other studies.'2 Cancer incidence data are available oniy during the subjects' years of active employment at Du Pont. Ill nesses that occur after termination from the Company, or during retirement, are not recorded in the registry. Using a computer program developed by Monson,'1 the observed number of cancer cases diagnosed during the period 1964 to 1977 was compared with the estimated expected number baser) on the experience of the non chemists and on the Third National Cancer Survey (TNCS) rates.Standardized incidence ratios (SI FIs), the ratios of the incidence rates of the chemists and the comparison group, standardized to the age and calendar year distribu tion of the person-years of the chemists, were calculated. Ninety-five percent confidence intervals were con structed, assuming a Poisson distribution of the observed number of events.'5 The TNCS rates were used to detect excess risks occur ring in both the chemist and the nonchemist cohorts which would not be apparent from internal comparisons. A third comparison group, all Du Pont salaried employees from each year of the study observation period, was considered. The more inclusive Du Pont salaried group was larger than the nonchemist group, but less comparable to the chemists with respect to year hired and socioeconomic status. The results based on all Du Pont salaried employees were similar to the nonchemist- based analysis and will not be presented. For all sites combined and for any site with three or more chemist cases, the authors examined the effects of age and decade hired, and usual and ever department af filiation. For each vaiue of the factors, the expected number of cases was generated by consideration of the nonchemists with the same value for the factor. The effects of alcoholism were also examined. A sub ject was classified as an alcoholic if he had a cause of death or disability due to alcoholic cirrhosis of the liver or to alcoholic psychoses, or had been identified in a company-wide survey of plant physicians conducted in 1964."" An AHI claim for alcoholism can be filed for an episode of illness or for a 30-day treatment course at an alcoholism rehabilitation center. Referral to a rehabilita tion center may be made by either the employee or by management. . Mortality. -- The chemist and nonchemist cohorts were followed to December 31, 1977, regardless of employment status. The mortality data were obtained from two sources, Du Pont records and the Social Securi ty Administration (SSA). The Medical Division maintains a mortality file which records all deaths among active employees and among pensioners. The file contains the life insurance claim data, supplemented by information 486 DBZ02008 Table 2. - Vital Status, as of December, 1977, of the Chemists and Nonchemist Cohorts, by Sex. Vital Status - Total Alive Active employees Pensioners ' Terminated from company, not pensioned Deceased Active employees Pensioners Terminated Iron company, not pensioned . Percent ol total lor the sex-cohort group tLess than 1 /< ' Clwmliu Mon Women 3,666 3,468(95)* 2,722(74) 448(12) 76 69(92) 41(55) 13(17) 316(9) 198(5) 111(3) 76(2) HI 15(20) 6(8) 4(5) 1(1) KD Nonchemlits Men Women 19,263 17,243(90) 10,867(56) 5,465(28) 673 653(97) 454(67) 170(25) 911(5) 2,020(10) 897(5) 1.071(6) 29(4) 20(3) 9(1) 10(1) 52T H ' from the employees' medical records and from the plant physicians. As of December 31, 1977, 187 male and 5 female chemists and 1,979 male and 19 female non chemists were known to have died, while actively employed or after having retired. Of the 329 male chemists who left the company before retirement or death, the SSA identified 260 as alive, 11 as deceased, and 58 as status unknown. Persons not known to be dead were assumed to be alive until the end of the study. One female chemist and one female and 51 male nonchemists were known to have died after termination from Du Pont (Table 2). The authors requested copies of death certificates from state departments of health, but did not try to ob tain certificates for 11 nonchemists and one chemist who died outside the United States. Certificates were obtained for 185 (93%) of the male and 6 (100%) of the female deceased chemists and for 1,863 (92%) of the male and 15 (79%) of the female deceased nonchemists. When a certificate was unobtainable, the cause of death from the life insurance claim was utilized. Ninety percent of the available certificates contained the same cause of death as the life insurance claim. Ninety-nine percent of the deaths attributed to cancer on the death certificate were also attributed to cancer on the life insurance claim. Cause of death was coded according to the Interna tional Classification of Diseases, Adapted for Use in the United States." " The Seventh Revision was used from Table 3. -- Observed and Expected Cencer Cases* Among Male Chemists, Aged 20-64, DiagnosedTn 1964-1977, Based on the Nonchemist Cohort Rates. Site (ICD)t Obs All sites (140-239) Buccal cavity (140-149) ' Stomach (151) Small and large Intestine (152-153) Rectum (154) Liver and biliary passages (155-156) Pancreas (157) Lung (162) Melanoma (172) Prostate (185) Bladder (186) Kidney (189) ' Brain (191) Endocrine glands, excluding thyroid (194) Lymphatic and hemalopoletic system (200-209) Lymphosarcoma (200) Hodgkin's disease (201) Ollier lymphomas (202) Multiple myeloma (203) Leukemia (204-207) ' Other neoplasms ol the lym- phopolellc system (20B)-209)1 All other sites 61 1 2 7 3 1 2 8 6 7 3 4 2 1 12 3 2 1 1 3 2 0 "Excluding non-melanollc skin cancer THighth Revision, international Classification ol Diseases'! ^Standardized incidence ratio (SIR) = (cOserved/expecled) x IDO 95'/i conlldence intervals'! ' IPolycythemia vera and adenogenic myeloid metaplasia Exp SIR* 86.5 71 2.1 48 2.4 83 11.6 60 4.4 . 68 0.9 . -- 1.1 181 20.0 40 8.3 97 4.5 . 156 2.0 150 3.7 108 ' 3.0 67 0.5 -- 10.0 120 2.T 142 1.6 127 1.5 67 0.9 -- 3.5 85 0.4 -- 12.0 -- C.l. ol SIR- 54-91 2-235 14-275 26-119 17-185 -- 30-598 19-76 45-184 68-308 38-407 34-261 11-221 -- 65-203 36-385 21-419 3-332 -- 22-231 ' -- --` 487 DBZ02008 93 Table 4. -- Observed and Expected Cancer Cases* Among Male Chemists, Aged 20-64 Years, Diagnosed In 1964-1977, Based on the Third National Cancer Survey, 1969-1971. 511b All sites - Buccal cavity Slomach Small and large Intestine Rectum Liver and biliary passages Pancreas Lung Mefanoma Prostate Bladder Kidney Brain Endocrine glands, excluding thyroid Lymphosarcoma Hodgkin's disease OthBr lymphomas Mulllple myeloma Leukemia All othBr sites Obs 61 1 2 7 3 1 2 8 ''8 7 3 4 2 1 3 2 1 1 3 2 Exp 112.0 8.5 3.8 9.9 5.7 1.6 3.7 28.4 3.3 5.6 6.6 3.8 3.4 0.3 2.7 2.1 1.0 1.2 3.3 17.1 SIRt 54 12 53 71 53 62 54 28 239 124 45 105 59 -- Ill 97 ' 97 81 91 -- Excluding non-melanotic skin cancer tStandardlzed Incidence ratio (SIR) (observed/expected) x 100 195% conlldence Intervals'5 C.l. of SIRT 42-70 1-5B 9-175 31-140 14-145 3-306 9-178 13-54 111-454 54-246 12-123 33-253 . 10-196 -- 28-303 16-322 5-479 4-397 23-249 Table 5. -- Observed and Expected Numbers of Deaths Due to Selected Causes Among Mate Chemists, 1964-1977, Based on (he Nonchemist Cohort. Cause (ICO)* All causes Malignant neoplasms (140-209) Stomach (151) Large intestine (153) Rectum (154) 'Liver and biliary passages (155-156) Pancreas (157) Lung (162) Proslate (105) Bladder (168) Kidney (1B9) Brain (191) Thyroid (193) Lymphatic and hematopoietic cancer (200-209) Lymphosarcoma and retlculosarcoma (200) ' Hodgkin's disease (201) Leukemia (204-207) Other lymphatic cancer (208-209) Diabetes melllius (250) Circulatory system diseases (390-45B) Arteriosclerotic heart disease (410-414) Cerebrovascular disease (430 - 438) Pneumonia (480-486) Gastric and duodenal ulcer (531-533) Cirrhosis ol the liver (571) Accidents (E800-E949) Suicide (E950-E959) Residual Obs 198 ... .43 ... 2 12 2 1 . '2 ... 9 2 0 2 1 1 9 1 2 4 2 .1 96 75 . 15 3 .1 '5 14 8 27 `Eighth Revision, International Classification ol Diseases' `fSIandardlZBd mortality ratio - (observed/expected) x 100 $95% confidence Intervals15 488 Exp 241.0 . 62.4 2,2 B.7 0.9 -1.7 1.7 21.9 2.8 1.1 2.3 3,0 0.2 7.5 1.2 0.7 3.7 2.0 3.2 128.1 103.3 10.8 1.1 0.6 5.7 13.5 5.6 20.8 . SMR} 82 69 90 178 1 " . 69 120 41 72 --` 86 33 -- 120 85 -- 109 ' 102 32 75 73 138 281 -- 88 104 144 -- C.l. OfSMfcf 71-94 51-92 .15-300 97-305 -- 3-290 . 20-389 20-75 12-236 . -- 15-287 2-164 -- 59-220 ' 4-411 -- 34-281 17-337 2-154 61-92 57-91 81-224 69-742 -- 32-194 . 59-17066-271 - DBZ02008 94 1964 through 1969, and the Eighth Revision, from 1969 through 1977, Through the use of the Monson program, the observed number of deaths among the chemists was compared with the estimated expected number based on the ex perience of the nonchemists and on the U.S. mortality rates. Standardized mortality ratios (SMRs) and 95% con fidence intervals were calculated in the manner described previously. An analysis based on all Du Pont salaried personnel employed during the study observation period yielded results similar to those of the nonchemist-based analysis and will not be presented. Proportional mortality ratios (PMRs) also were calcu lated, comparing the mortality patterns of the chemists with those of white male population of the United States. Results Cancer Incidence. -- While employed as chemists at Du Pont, the 3,686 men accrued 44,656.8 person-years of observation. With the exclusion of non-melanotic skin cancer, 61 cases of cancer were diagnosed during the period 1964 to 1977. The expected numbers of cancer of all sites were 86.5 (SIR = 71) and 112.0 (SIR = 54), based on the nonchemist and the TNCS rates, respectively. Table 3 shows the observed and the expected numbers, the SIRs, and the 95% confidence intervals of the SIRs, based on the nonchemist experience, for all combined and for selected cancer sites. Approximately one-half of the deficit in overall cancer morbidity was attributable to the low lung cancer morbidity. Slight excesses, based on small numbers, were seen for cancers of the pancreas, the prostate, the urinary bladder, and the lymphatic and hematopoietic system. A slight excess of prostatic cancer is seen among chemists ever and usually employed in basic chemicals and finishes (3 observed, 1.0 expected). All the chemists diagnosed with kidney cancer were employed in basic chemicals and finishes (even 4 observed, 0.6 expected; usual; 4 observed, 0.4 expected). Slight excesses were seen in the plastic resins and fibers chemists for lymphosar coma (ever: 2 observed, 0.3 expected; usual: 2 observed, 0.2 expected) and leukemia (even 3 observed, 1.1 ex pected; usual: 2 observed, 0.4 expected). The number of cases is too small for multiple breakdowns to be meaning ful. The comparison of the chemists with the TNCS popula tion was generally consistent with the nonchemist-based analysis (Table 4). However, melanoma was more com mon among the chemists than expected (8 cases ob served, 3.3 expected, SIR = 239), an association that does not appear in the nonchemist-based analysis. An excess of melanoma is observed in the nonchemists themselves in comparison with the TNCS population (38 cases ob served, 17.1 cases expected, SIR = 223). In the internal comparison with the nonchemists, the chemists do not appear to have excess melanoma risk. The female chemists accrued 747.0 person-years of observation. During the period 1964 to 1977, five cancers were diagnosed: one each of the breast, uterus, and ovary, one case of leukemia, and one cancer of unknown primary site. The expected number of cancers of all sites Table 6. -- Observed and Expected Number ol Deaths Due to Cancer of the Large Intestine Among Male Chemists, 1964-1977, by Work History Characteristics, Baited on the Nonchemist Cohort. Age hired 20-24 25-29 30-34 . 35-44 Decade hired 1920-1929 1930-1939 1940-1949 1950-1959 Usual department group Plastic resins and libers Basic chemicals and finishes Petrochemicals Elastomers Research, development and engineering Atomic energy Total . Obs Exp 3 2.1 4 3.2 4 0.6 1 0.4 0 1.1 4 2.3 4 1.2 4 2.1 3 3.4 4 2.0 2 0.3 0 0.2 2 0.5 1 0.0 12 6.7 was 2.7 (SIR = 185), based on both the nonchemist and the TNCS rates. Because of the small number of cases, no further analysis of the women's experience will be presented. Mortality. -- From 1964 to the earlier of either the d ate of death or the end of the study, the male chemists ac crued 50,538.7 person-years of observation and "198 deaths. The expected numbers of deaths were 241,0 (SA4R = 82) and 418.3 (SMR = 47), on the basis of the ex perience of the male nonchemists and of the U.S. white male population, respectively. In comparison with the experience of the nonchemists, deaths due to malignant neoplasia were less frequent than expected (43 observed, 62.4 expected, SMR = 69) (Table 5). An elevated SMR was seen for cancer of the large intestine (12 observed, 6.7 expected, SMR = 1783- A large deficit was observed for deaths due to carcinoma of the lung. Nine lung cancer deaths occurred, 21.9 were ex pected (SMR = 41). Slight excesses were observed for deaths due to Hodgkin's disease, to all lymphatic and hematopoietic cancer, and to cerebrovascular disease, pneumonia, and suicide. A low SMR was seen for arteriosclerotic heart disease. Table 6 presents the observed and expected numbers for cancer of the large intestine according to work history characteristics. Deaths occurred among chemists usually employed in plastic resins and fibers, basic chemicals and finishes, petrochemicals, research and engineering, and atomic energy. ' The associations seen in the total cohort were generally found in each department. Chemists usually employed in the elastomers department experienced slightly more deaths than expected (13 observed, 11.8 expected, S7*v1R = 110); chemists in the other departments had SMRs v^ell below 100. The elastomers department's relative excess was due to deaths from coronary heart disease and suicide. However, for most departments and most causes of death, the number of deaths was too small for any meaningful comparisons. 489 DBZ0200S 95 Table 7. -- Observed ind Expected Numbers of Deaths Duo to Selected Causes Among MalB Chemists, 1964-1977, ' Based on U.S. Mortality Rates. ClUII All causes Malignant neoplasms Stomach large Intestine Rectum ' Liver and biliary passages Pancreas Lung Prostale Bladder Kidney Brain Thyroid Lymphatic and hematopoietic system Lymphosarcoma and retlculo- sarcoma Hodgkin's disease Leukemia Other lymphatic cancer Diabetes mellltus Diseases ol nervous system and sensory organs Diseases ol the circulatory system Arteriosclerotic heart disease Cerebrovascular disease Respiratory diseases Pneumonia Digestive system diseases Gastric and duodenal ulcer Cirrhosis el tho liver External causes Accidents Suicide Residual - Obi - 198 43 2 12 2 1 2 9 2 0 2 1 1 9 1 2 4 2 1 2 SB 75 15 7 3 11 1 5 22 14 8 16 *SMR Standardized mortality ratio = (observed/expected) x 100 t95'/i confidence Intervals'* In comparison with the U.S. white male general population, 86.0 deaths due to cancer were expected, yielding an SMR of 50 (Table 7). An excess was observed for deaths due to cancer of the large intestine (12 ob served, 6.9 expected, SMR = 174). Two Hodgkin's disease deaths occurred; 1,3 were expected (SMR = 150). Four leukemia deaths occurred; 3.4 were expected (SMR = 119). Deficits are seen in numbers of deaths due to car cinoma of the lung (9 observed, 30.7 expected, SMR = 29), arteriosclerotic heart disease, respiratory and digestive system diseases, accidents, and suicide. The deaths and person-years of exposure were categorized by employment status. Active employees had the lowest SMRs, pensioners had the highest, and ter minated, nonpensioned employees had intermediate SMRs. The last group, the most difficult to trace, did not reveal any associations not seen in the active employees or the pensioner group. Risk of cancer of the large intes tine was elevated relative to the overall 5MR in both ac tive and pensioned employees, but particularly among the pensioners. Only the active employees had slight ex cesses in deaths due to cancer of the kidney and to Hodg kin's disease. Only the pensioners had excesses in deaths due to prostatic cancer, cerebrovascular disease, diseases of the digestive system, cirrhosis of the liver, and suicide (Table 8). 490 . Exp 418.3 86.0 3.6 6.9 2.4 1.3 4.7 30.7 2.8 1.9 2.4 3.5 0.2 9.1 2.2 1.3 3.4 2.1 5.8 3.9 198.9 151.5 20.0 20.9 7.3 26.5 2.5 17.1 54.2 34.7 13.6 22.1 SMR* 47 50 55 174 85 78. 43 '29 70 -- 83 29 -- 99 46 150 119 94 17 51 48 50 75 33 41 41 40 29 41 40 59 C.l. ofSMRj 41-54 37-660 9-1B4 94-296 14-275 4-379 7-141 14-54 12-236 -- 14-275 1-141 -- 48-182 2-224 26-508 37-284 16-315 1-85 9-169 39-59 39-62 44-121 15-66 11-112 22-72 2-193 11-65 26-60 23-66 27-112 In the proportional mortality analysis, the method usecS in several of the previous chemist studies, deaths due to malignant neoplasia were found to be more frequent than expected (Table 9). A larger than expected proportion o~f deaths among active employees was due to lung cancer, but deficits were seen among the pensioned and other ter minated employees: Colon cancer, arteriosclerotic heart disease, and cerebrovascular disease were responsible for larger proportions of deaths than expected. Deaths at tributable to prostatic cancer occurred more frequently than expected among pensioners. The risk of deaths from accidents was lower than expected, and,.: among ter minated employees, suicide accounted for a larger than expected proportion of deaths. Among the female chemists, six deaths occurred during 994.7 person-years of observation. The expected numbers were 1.8 (SMR = 327) and 4.8 (SMR = 125), on the basi-s of the experience of the female nonchemists and of the U.S. white female population, respectively. The causes oi death were lung cancer, leukemia, abscess of the lung, house fire, and in two cases, suicide. Compared to' the U.S. rates, only'0.1 deaths due to suicide were expected- Discussion The Du Pont male chemists experienced lower overal I mortality and lower cancer morbidity and mortality them DBZ0200S 96 Table 8. -- Observed and Expected Numbers ol Deaths Due to Selected Causes Among Male Chemists, 1954-1977, by Employment Status, Based on U.S. Mortality Rales. Causa All causes Malignant neoplasms LargB Intestine Lung Prostate Kidney Lymphatic and hematopoietic cancer Hodgkin's disease Circulatory system diseases Arteriosclerotic heart disease Cerebrovascular disease Respiratory diseases Digestive system diseases Cirrhosis of the liver Suicide Residual . Actives Obs Exp SMR* ill 309.7 . 36 25 61.8 40 5 4.7 106 7 22.0 32 0 1.2 ` -- 2 1.8 110 6 7.0 86 2 1.2 174 49 141.0 35 , 40 108.7 37 5 12.6 40 2 13.5 15 7 21.6 32 3 14.4 21 4 11.9 34 24 59.9 *SMR Standardized mortality ratio = (observed/expected) x 100 Pensioners Obs Exp SMR 75 91.5 83 16 21.0 76 6 2.0 '307 2 7.5 27 2 1.6 125 0 0.5 -- 3 0 1.7 0.1 1_72 42 50.7 83 31 37.2 83 9 6.8 133 6 6.8 74 4 3.6 111 2 1.8 114 2 0.9 7 8.5 - expected on the basis of the experience of the industrial and general population comparison groups. Deaths due to cancer of the large intestine and to cerebrovascular disease occurred more frequently than expected. Small excesses of cases of prostatic cancer and of melanoma were observed. Sharply decreased rates of lung cancer in cidence and mortality and of arteriosclerotic heart disease mortality were observed. The female chemists ex perienced greater than expected mortality from all causes combined and from suicide. Among the men, the finding of excess mortality from cancer of the large intestine is based on 12 observed deaths, a number larger than that for any other cancer site. Relative to the experience of the U.S. population, the excess of colon cancer was highest among pensioners. The deaths were distributed throughout all but two major Du Pont department groups. Larger than expected proportions of deaths due to cancer of the intestine and the rectum occurred among deceased AC5 and British RIC members and among chemists dying in the state of California during the years 1959 to 1961.'411 Among the Swedish chemistry graduates the number of deaths due to malignancies of the digestive system was approximately that which was ex pected.4 A recent update of the Swedish cohort showed a slight deficit of digestive system cancer, deaths." Among chemists dying in the state of Washington there was one death due to cancer of the large intestine; two deaths were expected.1 The incidence of cancer of the large intestine was not elevated with respect to the comparison groups. However, the incidence data are limited to active employees who are under age 65. If the excess frequ ency appears after age 64, as suggested by the U.S.-based mor tality analysis, no excess in incidence would be expected. The Du Pont chemists show an excess of prostatic cancer relative to nonchemist salaried employees. The occurrence of prostatic cancer increases directly with age,'4 so the Du Pont incidence data, restricted to cases diagnosed before the age of 65, may underestimate the Table 9. -- Observed and Expected Numbers and Proportional Mortality Ratios for Deaths Due to Selected Causes Among Male Chemists, 1964-1977, by Employment Status, Based on U.S. Proportional Mortality Bates. Ciuie All causes Malignant neoplasms Large Inlesllne Lung Prostate Kidney Lymphatic and hematopoietic cancer Hodgkin's disease Leukemia Arteriosclerotic heart disease Cerebrovascular disease Respiratory diseases Digestive system diseases Cirrhosis ol the liver Accidenls suicide Actives Obi Exp PMB* 111 111.0 100 25 18.3 137 5 1.3 377 7 5.6 125 0 0.2 -- 2 0.5 -- 6 2.8 213 2 0.6 -- 3 1.1 284 40 29.8 134 5 3.8 132 2 3.9 51 7 8.2 85 3 5.6 54 12 19.2 62 4 7.1 56 *PMR Proportional mortality ratio * (observed/expBded) x 100 Pensioners .Obs Exp PMR 76 76.0 100 16 17.9 89 6 1.5 398 2 6.9 29 2 0.8 -- 0 0.5 -- 3 1.5 196 0 0.1 -- 1 0.6 -- 31 30.4 102 9 4.1 221 5 4.6 109 4 4.0 100 2 2.4 83 1 3.0 34 2 1.3 15B Tom Obs Exp PMR 19B 198.0 100 43 37.2 116 12 2.8 426 9 12.7 71 2 0.6 -- 2 1.1 187 9 4.7 192 2 0.9 4 1.7 230 75 61.9 121 15 7.5 201 7 8.0 87 11 13.9 79 . 5 9.4 53 14 25.8 54 8 9.9 81 . 491 DBZ0200S 97 Table 10. -- Proportional Mortality Ratios and Estimated Standardized Mortality Ratios for Selected Causes of Death Among Male Chemists. Cause All cancers Large Intestine Lung Leukemia Suicide Du Pont U at at' (Overall SMR - 47) < 65 Years PM>W SMR^t PMIW SMR,,,t 116 50 125 426 174 115 71 29 116 230 119 132 81 59 117 59 54 54 62 55 U et al' i.65 Years PMRrt, SMRU| 115 54 134 53 BO 38 178 84 100 47 Mlltiam' Washington St. PMR^, SMRU1 99 47 43 20 63 30 213 100 195 92 Petersen et at* Registrar General' California St. England and Wales PMH*, SMRU1 PMR*,, SMR,,, 91 43 96 45 207 97 M/A N/A 64 30 63 30 116 55 320 150 248 117 378 178 PMRWS - Cause-speclllc proportional mortality ratio tSMRSM Cause-speclllc standardized mortality ratio Cause-speclllc estimated standardized mortality ration* PMRy,, x overall SMR (Irom Ou Font data)22 N/A * Data not available In published report risk. Both of the chemists who died from prostatic cancer were pensioners, again suggesting that the greatest risk is in the oldest age groups, for whom incidence data are unavailable. The problem would be compounded if undetected cancer were associated with retirement or with employment termination. An excess of prostatic cancer is also seen among the Swedish chemistry graduates and among the Californian chemists. The incidence of melanoma was greater than expected among the chemists and the nonchemists, based on the experience of the TNCS population. A similar excess was seen among all Du Pont employees in the period 1956 to T974.'3 A possible explanation may be occupational ex posure to chemicals, some of which are suspected of causing non-melanotic skin cancer." Another putative cause is exposure to solar radiation. The majority of Du Pont plants are located in the southeastern United States where exposure to the sun is high. Fifteen southern states have the highest mortality rates for melanoma, account ing for 37% of the melanoma deaths In the United States." Sixty-five percent of the Du Pont salaried em ployees are located in these same states. Excess deaths due to cerebrovascular disease were seen in the Du Pont chemist cohort and among chemists dying in the state of Washington during 1950 to 1971.3 Among the men and women, slightly more deaths than expected were due to suicide. Reports from the ACS, from the states of Washington and California, and from the Registrar General of England and Wales also mentioned excess numbers of deaths due to suicide.3 3 *7 An excess of suicide among female physicians suggests that a sociological phenomenon may account for the excess among female chemists.33 Other excesses were anticipated on the basis of the previous studies of chemist populations. Almost all of the previous studies reported excess deaths due to all sites of cancer combined and to lymphatic and hematopoietic malignancies, Several studies reported excesses in deaths due to bladder and pancreatic cancers. There are several possible reasons why the Du Pont ex perience was not consistent with the earlier reports of ex cesses, First, the outcome measures differed: thls.study is based on cancer incidence and absolute mortality, while most of the earlier studies were based on proportional mortality. DecouflA et al33 33 recently discussed the bias of. PMR analyses when the overall death rate differs from 492 that of the comparison group. Low overall mortality can cause PMRs to overstate risk. For deaths due to cancer, the Du Pont male chemists had an SMR of 50 and a PMR of 116, relative to the appropriate U.S. comparison population. The earlier chemist populations, with excess proportional cancer mortality also may have had low ab solute cancer mortality. Table 10 presents the observed PMRs and the estimated SMRs for selected causes of death from the previous chemist cohorts and from the Du Pont chemists. The SMRs were estimated by multiplying each cause-specific PMR by the Du Pont overall SMR. Few of the elevated PMRs are converted into elevated SMRs. In each study, the observed or estimated SMR for leukemia is greater than the SMR for deaths due to all cancers. Conversely, the observed or estimated SMR for lung cancer is smaller than the SMR for all cancers. The Swedish absolute mortality study* showed greater cancer mortality than expected, inconsistent with the Du Pont results. Second, the follow-up period may be too short to per mit detection of the anticipated excesses. Slight excesses of deaths due to lymphatic and hematopoietic malignan-' cies and of cases of bladder and of pancreatic cancer were seen, but the observed numbers of events rarely ex ceeded the expected numbers by more than one. Further follow-up and reanaiysis may yield enough cases to con- - firm the earlier reports of increased risk of these diseases among chemists. ` The Du Pont subjects were hired as early as the 1920s, were actively employed in 1959, and were followed until December 31, 1977. The length of follow-up, 19 years, was thought to be adequate for investigation of lym phatic and hematopoietic diseases; the average incuba tion period for leukemia associated with benzene ex - posure has been estimated at between 6 and 14 years.35 The number of persons in the cohort was greater than in the Swedish study, the other absolute mortality study of chemists. However, the chemists were younger and the events fewer than estimated. Approximately three-fourths of the chemists were active employees at the close of the study. The numbers of cases and deaths are small. Third, associations may have been obscured by expo sure misclassification. The Du Pont cohort was restricted to persons employed as chemists who were thought to have had more exposures than either the previously studied chemist cohorts or the nonchemists. However, if DBZ0200S 98 only a small subgroup of the chemists were exposed to hazardous substances, the excess risk would be diluted by the larger unexposed group at low risk. Also, some of the Du Pont nonchemists sustained exposure to hazardous chemicals, contributing to the misclassification problem. Finally, the cancer incidence data are limited to cases diagnosed before employment termination. No cancers detected after age 65, the mandatory retirement age dur ing the observation period, are identified. This restriction reduces the study's ability to detect excesses of disease increasing directly with age. Large deficits in overall mortality and in cancer mor bidity and mortality rates were observed among- the chemists. Deficits were anticipated in the comparisons with the U.S, general population because of the "healthy worker effect,"" but not In the comparisons with the Du Pont nonchemist group. Selection bias is an unlikely explanation for the defi cits. The chemist and nonchemist subjects were selected in the same manner from preexistent records that were created without regard for future disease or death. Observation bias is also unlikely although incomplete ascertainment of deaths may have occurred. Deaths and cancer cases were identified through the Du Pont epidemiologic surveillance system which was established for all employees prior to the initiation of this cohort study. The SSA identified deaths among the employees who terminated employment prior to retirement. Subjects were classified as dead, alive, or status unknown. The sub jects not known to be dead were assumed to be alive un til the end of the study. If all the chemists whose status was unknown had been deceased, but all such non chemists were alive, the observed number of deaths would increase by 58 to 256. The expected number would decrease slightly because of fewer chemist person-years of observation, being at most 241.0, yielding an SMR of 106. It is unlikely that this extreme situation occurred. More than half of the chemists whose status was unknown and who had been last employed by Du Pont in the Delaware area had private or business phones listed in their names In the local 1978 phone books. The SSA also may have erred In classifying as alive people who were deceased. In previous studies as many as 6% of persons known to be dead were-classified as alive by the SSA." The most likely explanation for the deficits in the mor bidity and mortality experience of the chemists is con founding. Factors associated with cancer may be distributed unequally between the chemist and the com parison groups. The significant deficit in lung cancer mor bidity and mortality among the chemists suggests smok ing habits different from those of the comparison groups. The chemists also had low mortality from cardiovascular disease and low incidence for several other cancers thought to be related to smoking: cancer of the buccal cavity, esophagus, stomach, and larynx.1* " The chemists had only one more case than expected of cancers of the bladder and of the pancreas, which also are associated with cigarette smoking. - It was not possible to obtain smoking histories for past or present Du Pont chemists. Questionnaires and inter views could not be administered, and the. smoking histories recorded in the annual medical examinations are unreliable. However, some information on smoking is known. The Du Pont chemists are prohibited from smok ing in the research laboratories, while persons in many of the nonchemist occupations, such as lawyers and accoun tants, have the freedom to smoke during the workday. On-the-job smoking behavior may influence smoking habits outside of work. Data on smoking and occupation from the National Center for Health Statistics 1970 Household Interview Survey showed that the smoking habits of chemists differed from those of other profes sional and technical workers and from those of other business management employees, such as accountants and auditors.10 Only 29% of chemists are current smokers, whiie 33% of accountants and auditors. 36% of professional, technical and kindred workers, and 37% of personnel and labor relations workers are current smokers. The percentage of smokers who smoked one pack or more per day was 59% for the chemists, 66% for professional, technical and kindred workers, 69% for ac countants and auditors, and 71 % for personnel and labor relations workers. A recent National Health Survey31 stated that 41% of managers and administrators were current smokers while only 30% of professionals, such as chemists, presently smoked. The deficits observed in mortality from all causes, from all cancers, and from lung cancer are compatible with the differences in smoking habits reported in the surveys, but there may be other confounding factors also. For exam ple, the chemists probably have a higher socioeconomic status than the comparison groups. The nonchemists in clude persons, such as foremen, who began employment on the hourly wage roll. , Summary In this study, Du Pont male employees with the job ti tle chemist or with associated titles experienced lower rates of overall mortality and of overall cancer morbid ity and mortality than other salaried employees of the com pany. A large deficit is seen in lung cancer and arteriosclerotic heart disease. The chemists appear to be at slightly higher risk of death from malignancies of the colon and from cerebrovascular disease. The incidence rates of melanoma and of cancer of the prostate-are slightly higher than expected, relative to the TNC5 and the experience of nonchemists, respectively. Among female chemists, deaths from all causes combined and from suicide occurred more frequently than expected. The low overall mortality, among the male chemists resulted in a larger than expected proportion of dea ths due to cancer. The cancer excesses in the previous pro portional mortality reports on chemists may also reflect low absolute mortality and not absolute excesses. Anticipated excesses of certain types of cancers were not observed, possibly because of the use of absolute mortality rates, inadequate length of follow-up, exposure to hazardous chemicals by the referent group, or restric tion of case identification to active employees. Deficits in overall mortality rates and in lung cancer rates were observed. These findings may have been cdue to lower prevalence of smoking among the chemists, or to disparate socioeconomic status or other factors. 493 DBZ0200S 99 The authors wish to thank Dr. Bruce W. Karrh, Corporate Medical Director of Du Pont for his support of the study, Drs. Brian MacMahon, Richard R. Monson, and David H. Wegman for epidemiologic advice, Dr. Alan S. Morrison for his review of the manuscript. Dr. Vann C. Brewster for his administrative assistance, and Audrey E. L. Gidney for her programming assistance. William E. Payerweather, Brian H, Wright, and George W. Zervas also provided program ming assistance, Ann D. Monson was the nosologlst Clerical duties were pen formed by Mary Donahue, Louise Mowbray, Gerrl A. Nardo, Lisa P. Ranken. and 15. Miettlnen OS: Comment I Am Stat Assoc 69:3BO-3B2,1974. 16. Pell S and D'Alonzo DA: A five-year mortality study of alcoholics. I Occup Med 15:120-125,1973. 17. International Classification of Diseases, Adapted for Indexing Hospital Records by Disease and Operations, U.S. DHEW, Public Health Service Publ, No 62-719. Washington, D.C.: U.S. Covt. Print. Office, 1962. Delores Saunders. Kathryn C. Buchanan, Doris Carmo, Gertrude M Crowley, 18. Eighth Revision International Classification of Diseases: Martha DiOomenicis, and Mary CyLoughlin provided secretarial services. Adapted for Use In the United States. U.S. DHEW, Public Health Ser vice Publ. No. 67-1693. Washington, D.C.: U.S. Govt. Print Office, References 1967. 1. LI FP, Fraumenl JF Jr, Mantel N, et al: Cancer mortality among 19. Olln GR and Ahlbom A: The cancer mortality among Swedish chemists. / Nat/ Cancer Inst 43:1159-1164,1969. chemists graduated during three decades. Environ Res 22:154-161. 2. U FP: Suicide among chemists. Arch Environ Health 19:518-520, 1980. 1969. 20. Emmett EA: Occupational skin cancer: A review. 1 Occup Med 3. Mllham S Jr: Occupational Mortality in Washington State, 17:44-49,1975. 1950-1971, U.S. DHEW, Public Health Service Publ. No. 76-175. 21. Mason TJ, McKay FW, Hoover R, et al: Atlas of Cancer Mortali Washington, D.C.: U.S. Govt. Print. Office, 1976. ty for U.S. Counties: 1950-1969. U.S. DHEW, Public Health Service 4. Petersen CR and Milham 5 )r. Occupational Mortality in the Publ. No. 75-780. Washington, D.C.: U.S. Govt Print. Office, 1975. State of California, 1959-1961. U.S. DHEW, Public Health Service 22. Finseth K: Suicide among women physicians. JAMA 237:1693, Publ. No, 80-104. Washington, D.C.: U.S. Govt Print Office, 1980. 1977. 5. Olln GR; Leukemia and Hodgkin's disease among Swedish 23. Decoufld P, Thomas TL', and Pickle LW: Comparison of the pro chemistry graduates. Lancet 7991:916,1976. 6. Olin CR: The hazards of a chemical laboratory environment -- A portionate mortality ratio and standardized mortality ratio risk measures. Am J Epidemiol 111:263-269,1980. study of the mortality In two cohorts of Swedish chemists. Am tnd Hyg 24. Decoufle P and Thomas TL: A methodological investigation of Assoc ) 39:557-562,1975. fatal disease risks in a large industrial cohort. ] Occup Med 7. Occupational Mortality: The Registrar General's Decennial Sup 21:107-110,1979. plement for England and Wales, 1970-1972. London: Her Majesty's 25. Cole P and Goldman MB: Occupation, in Persons at High Risk Stationary Office, Office of Population Censuses and Surveys. 1978. of Cancer: An Approach to Cancer Etiology and Control, J. F. 8. Study unveils British scientists' way of death. New Scientist B0;93, Fraumeni, Jr. (Ed.). New York: Academic Press. 1975. 1978. 26. McMichael AJ: Standardized mortality ratios and the "healthy 9. Curalnlck L: Mortality by occupation level and cause of death worker effect": Scratching beneath the surface. I Occup Med among men 20 to 64 years of age, United States, 1950. Vital Statistics, 16:165-168,1976. Special Reports 53:141,1963. 27. OttMG, Holder BI3, and Langner RR: Determinants of mortality 10. Kupper LL, McMichael AJ, Symons MJ, et al: On the utility of . in an Industrial population. / Occup Med 18:171-177,1976. proportional mortality analysis. 1 Chron D/s 31:15-22,1978. 11. Annual Report, E. I. du Pont de Nemours and Company, Inc. 28. Kahn HA: The Dorn' study of smoking and mortality among United States veterans: Report on eight and one-half years of observa Wilmington, Del,: E. I. du Pont de Nemours and Company, Inc., tion. Natl Cancer Inst Monogr 19:1-125,1966. 1956-1977. 29. Hammond EC: Smoking In relation to death rates of one million 12. Pell S, O'Berg MT, and Karrh BW: Cancer epidemiologic men and women. Natl Cancer Inst Monogr 19:127-204.1966. surveillance In the Du Pont Company. J Occup Med 20:725-740,1978. 30. Sterling TD and Welnkam II: Smoking characteristics by type ol 13. Monson RR: Analysis of relative survival and proportional mor employment. / Occup Med 18:743-754,1976. tality. Compul Biomed Res 7:325-332,1974. 31. Use Habits Among Adults of Cigarettes, Coffee. Aspirin, and ,14. Cutler S) and Young )L |r (Eds.): Third National Cancer Survey: Sleeping Pills, U.S. DHEW, Public Health Service Publ. No. 60-155^. Incidence data. Natl Cancer Inst Monogr 41:1-454, 1975. Washington, D.C: U.S. Covt Print Office, 1980. 494 DBZ0200& 00