Document QMLxRx66VQNQD835VQ5okoQLo

A Retrospective Mortality Study of L Workers in T h r e e M a j o r U.S. Refineries and Chemical Plants Part 1 : Comparisons with U.S. Population Nancy M. Hank, Ph.D.; Leba G. Shallenberger, M.P.A.; Donna L. Donaleski; and Eugene A. Sales, M.S. nr) A dynam' retrospective cohort study was performed to ( ex%e'i the mortality experience o f 27,698 workers a t Exxon's refineries andchemical plants in Baton Rouge, la.; Baytown, Tex.; and BaywayfBayonne, NJ. Included were 15,437 regular employees who worked at least one month during the period)an. 1, 7970, through D e . 31, 1977,a n d 6,267 retirees who were alive as of )an. 7, 7970. There were 137,702 person-yean of observation. Mortality in this total studypopulation was generally lower than that of the US. population. Study follow-up was complete for 98.7% o f the study population. The standardized mortality ratio for the 3,798 deaths was 97, while that for deaths from all cancers (N=666)was 94. Certain slightly elevated disease- specific mortality ratios, although not statistically signifi- cant, could b e of bioiogical importance and merit further review. Cohorts of workers from three Exxon refineries and chemical plants in the United States were selected for mortality studies for the period 1970 through 1977. The study included employees at three plant sites, located in Louisiana, Texas, and New Jersey. The primary purpose was to compare the observed mortality in each plant population with that of the U.S. pop- ulation of similar age, sex, and race. Results of the study of the Louisiana plant cohort were published in 1982.' The three worker populations have been combined into a single cohort to address the following objectives: e.&om the Exxon Corporation, East N.I. iOr. aleski and Mr; Sales); and the Exxon Company, U.S.A., Houston Shallenberger)& AQtlress correspondence to Exxon Corporation, P.O. Box 235, Met- tlers Road, East Millstone, NI 00873 (Or.Hank). Journal of Occupational MedicineNol. 27, No. 41Apnl 1985 1. To compare the mortality experience of the total study population with that of the U.S. population. 2. To develop an internal comparison group for comparing subgroups within the cohort. 3. To compare mortality rates (directly adjusted for age, race, sex, and calendar year) among the three plant cohorts using the internal comparison group as the standard. 4. To compare mortality rates in the total study popu- lation by race, occupational categories, and smoking history. This report presents the results of the mortality comparison of this total study cohort and the US. population (objective one). Results relating to the remaining objectives will be dealt with in another report. Materials and Methods Study Design and Cohort Oefinition - A retrospective cohort study design was used to examine the mortality experience of workers at three Exxon US. refineries and chemical plants. This populationwas dynamic, in that em-. ployees entered the 1970 through 1977 observation period at various times and remained for varying periods. Mortality due to all causes and to selected specific causes was explored. The total study population (N=21,698) included 15,437 full-time regular employees with at least one month of service at any of the three US. plant sites (Baton Rouge, La.; Baytown, Tex.; Baywayfflayonne, N.J.) duringthe eightyear period Jan. 1, 1970, through Dec. 31, 1977. All 6,261 retirees who were alive as of Jan. 1, 1970, were also included in the population at risk. Employees who worked at more than one plant site were given credit for the years spent at each plant in the mortality analyses but were only counted once in each of the total population description 283 A --L,,L, according to the eighth revision of the International Classification of Discascs fICLIAJ.a In '25 instanccs where death certificates were unavail&le, information from death notices or company medical records was used. The U.S. cause-specific mortality rates by age. sex, race, and calendar year were used for comparison with the mortality experience of the total -study population. Analytical Methods Standardized mortalily ratios (SMRs) were uscd for comparing the mortality experience of the total study population with that of the U.S. population.'The following method was used to calculate the SMRs as percents:. -SMR = 0 x 100 E Rescription of study population. tables. There were 60 of these "multiple date of entry" -employees. Data Collection and Handling Extracts from company personnel and medical records provided Social Security number, company identification number, name, date of birth, sex, race, date of employment, date of termination, and status of last observation for each member of the study population. Date of death and underlying cause of death were extracted from death certificates. The under- lying cause of death was coded by a trained nosologist where 0 indicates the observed number of deaths in the study population for a specific cause of death: and E, the expected number of deaths in the study population, calculated by multiplying the age-, sex-, race-, calendar . year-, and cause-specific mortality rates of the standard U.S. population by the corresponding number of personyears at risk in the study population. Person-years at risk equals the sum of the total years of observation for mem- bers of the study population as they pass through each five-year age group, each five-year calendar period, and each plant. This i s sometimes referred to as a modified life-table approach.. SMRs calculated for different study populations are not I -Table 1 Oistribution of Total Study Population by Age, Race, and Sex at Entry Into Cohort ~ ~~ White' Black ~ Male Female Male Female Total Age* YI NO. y. NO. 78 No. YO No. % No. % c20 20-24 25-29 237 2.657 1,971 1.3 14.4 10.7 74 6.6 60 3.0 19 12.7 390 1.8 333 29.7 565 27.9 65 43.3 3.620 16.7 130 11-6 333 16.4 40 26.7 2,474 11.4 30-34 699 3.8 79 7.1 ' 98 4.8 13 35-39 388 2.1 51 4.6 31 1.5 5 40-44 845 4.6 83 7.4 28 I .4 3 45-49 1,643 8.9 98 8.8 66 3.3 1 50-54 1,985 10.8 * 72 6.4 139 6.9 0 55-59 2.097 11.4 63 5.6 190 9.4 0 60-64 1,889 10.3 60 5.4 176 8.7 2 65-69 1.589 8.6 40 3.6 136 6.7 0 70-74 1,164 6.3 . 17 1.5 96 4.7 1 75-79 783 4.3 80-84 341 1.9 85+ 114 0.6 12 1.1 6 0.5 2 0.2 61 3.0 30. 1.5 17 0.8 1 0 0 8.7 889 4.1 3.3 475 2.2 2.0 959 4.4 0.7 1ma 8.3 0.0 2.196 10.1 0.0 2.350 10.8 1.3 2.127 9.8 0.0 1,765 8.1 0.7 1,278 5.9 0.7 857 3.9 0.0 377 1.7 0.0 133 0.6 Total 18,402 100.0 1,120 100.1 2,026 100.0 150 100.1 n . 6 9 8 99.8 % of l Total 85 5 9 1 100.0 'Includes minorities other thin black 204 Mortality StudyManis et ai -Table 2 Distribution of Total Study Popuiatiori Iiy Year of Employment Employee' ' Employment 1900-1909 1910-1919 1920-1929 1930-1939 1940-1949 1950-1959 1960-1969 1970-1977 190 1,295 2.908 2.307 6.177 824 3.104 4.893 0.9 151 4.7 6.0 727 22.7 13.4 1.046 32.7 10.6 404 12.6 28.5 796 24.9 3.8 25 0.8 -14.3 22.6 32 1.o -17 0.5 Total 21.698 100.1 3,198 99.9 Employees hired during specific employment period t Deaths of employees hired during specific employment period directly comparable with one another since the computation involves an indirect adjustment p r ~ c e d u r e .A~ .r~esult was considered for further discussion i f (1)i t s SMR was based on five or more deaths and exceeded 100 andl or (2) i t s SMR was statistically significant at pc.05. Results -Description of the Study Population As shown in the Figure, of the total population of 21,698 employees, 42% were alive and still employed at the end of the observation period, 46% had retired (of these, 29% died during the study period), 10% had terminated employment before retirement, and 1%had died while still employed. Ninety- one percent of the total deaths (N=3,198) occurred among retirees. Follow-up for vital status was successful for all bot 284 (1%) of the study subjects. The 21,698 employees r-- I -Table 3 Distributionat Deaths by Year, 1970 Through 1977 Year of Death No. yo 1970 401 12.5 1971 396 12.4 1972 425 13.3 1973 442 13.8 1974 391 12.2 1975 382 11.9 1976 370 11.6 1977 391 -12.2 Total 3.198 99.9 1 I contributed 137,702 person-years for the eight-year observation period. At date of entry into the cohort, white males accounted for 85% of the total population, white females for 5%, black males for 9%, and black females for 1% (Table 1). There was a bimodal age distribution among white males, with peaks at ages 20 through 24 and 55 through 59. In the younger age groups there were higher proportions of females and black males than of white males. Approximately 59% of the total population was employed prior to 1950; this group contributed 98% of the deaths. Twenty-three percent of the employees began employment duringthe eight-year observation period CTable 2). Deaths were evenly distributed throughout the eight years (Table 3). Death certificates were available for all but 0.8% (N=25)of the study population deaths (N=3,198). Sixty-one percent of all deaths were due to diseases of -Table 4 Olstribution of Deaths by Major Cause, 1970 Through 1977 r Major Cause ut Oealh' NO. YO U.S. r o t Infective/pamitic diseases (000-136) All neoplasms (140-239) Endocrine/nutritiona~~e~odliisceases (240-279) Diseases of bloodmlood-forming organs (280-289) Mental disorders (290-315) Diseases of nenrOuS systemlsense organs (320-389) Diseases of circulatory system (390-458) Diseases of respiratory system (460-519) Diseases of digestive system (520-577) 'Diseases of genitourinary system (580-629) Diseases of skin/subcutaneous tissue (680-709) Oiseases of musculoskeletal system comedive tissue (710-738) Symptoms/senility/ilI-defined conditions (780-796) All external causes (8004999) 20 675 59 9 6 15 1,946 164 93 51 1 6 42 -111 0.6 21.1 1.8 0.3 0.2 0.5 60.9 5.1 2.9 -1.6 0.2 -1.3 3.5 0.6 19.3 2.3 0.3 0.5 0.9 52.2 5.9 3.7 1.3 0.1 0.3 1.7 -8.3 Total . L 3,198 100.0 97.6 I Journal of Occupational MedicineNol. 27, Na. 4IAprll 1985 285 r i , -Table 5 Distribution of Oeaths Due to Malignant Neoplasms by Major Site. 1970 Through 1977 Cancer Oeath Site' No. % , U.S. O/ot Buccal cavitylpharynx (140-149) All digestive organs (150-159) All respiratory system (160-163) Bonelconnectivetissuelskinlbreast (170-174) All genitourinary organs (180-189) Other and unspecified (190-199) All lymphopoietic (200-209) 14 2.1 196 29.4 217 ' 32.6 17 2.6 100 15.0 - -67 10.1 55 8.3 2.2 27.2 23.7 11.2 16.3 9.9 -9.5 i ~ _. Total 666 100.1 100.0 'ICDA. eighth revision codes shown in parentheses t Percent based on proportion of cancer category in relation to number of total cancer deaths' the circulatory system and 21% were due to malignant neoplasms (Table 4). Deaths of unknown cause (1% of total deaths) were included in the symptoms/senility/illdefined category. Table 5 shows the distribution of sitespecific cancer deaths. One third of the cancer deaths were due to respiratory system neoplasms, while 29% were due to cancers of the digestive system. In the 1975 US. population, 52% of all deaths were due to diseases of the circulatory system and 19% were due to malignant neoplasms; of these malignant neoplasms, 24% were in the respiratory system and 27% were cancers of the digestive system.' Comparison of the Mortality Experienceof the Total Study Population with That of the US. Population -Thenumbers of deaths due to all causes (SMR =90, to mental disorders (SMR=40), to diseases of the respiratory system (SMR=64), to diseases of the digestive system (SMR=71), and to all external causes (SMR =%I were significantly lower @<.OS) than expected. Deaths due to benigdunspecified neoDlasms (SMR- 1321 and to diseases of blood- hlo-o-d--fo-r-m---ing organs (SMR=l20) were greater than exDected&rt , the increases were not statistically significant (Table hl / The n u m b s o t observed deaths for all maliinant neopiasms was less than expected (SMR=W), b u t t h e difference was not statistically significant. Other cancer sites of interest for which numbers ofobserved deaths were less than expected included lung (SMR=SS), skin (SMR=99), genitourinary organs (SMR=83), leukemia (SMR=77), and all lymphopoietic cancers (SMR=89). SMRs for six sites were greater than 100, based on fne or more deaths, but none were statistically significant. Rateswere elevated for cancers of the tiver/gallMadder/bile ducts (SMR= lU), I -Table 6 StandardizedMortality Ratios (SMRs) for Selected. Major Causes of Oeath, 1970'Thtwgh1977t Observedl Cause of Death$ Expected5 SMR CI' All causes of death (000-999) Infective and parasitic diseases (000-136) . 3,198t3.515.2 2W24.0 91 88-94 83 51-128 Malignant neoplasms (140-209) 66wo5.0 94 87-102 1 BenigrVunspecified neoplasms (210-239) EndocrineInutritionaUmetabolicdiseases (240-279) Diseasesof blood/blood-formino orqans (280-289) . 916.8 132 61-252 59163.9 92 70.-~119 W7 5 120 G-936 Mental disorders (290-315) Diseases ot nemus systemhemeorgans (320/389) 6114.9 W24.0 40 15-88 63 35-103 Diseases of amlatory system (390-458) Oiseases of respiratory system (460-519) D i e s of digestive system (520-577) 1,946l1.980.8 164m4.4 93fl30.5 98 94-103 64 55-75 71 58-87 Oiseases of genitourinary system (580-629) Diseasesof musculoskeletal systmwnnedive 51t50.5 101 75-133 66.8 103 38-115 Tissue (710-738) Symptoms/senilityTlI-definedconditions (780-796) All external causes (ESWE999) 42t44.2 1111197.2 95 67-126 56 46-68 SMRs were not includedwhen the observed number of deaths was less than five (disaases of skiWsubwtaneoustissue [680-709] had only one death) t total: person-years = 137,702.5; number of employees = 21,698: number of deaths = 3,198 3 ICDA. eighth revision codes shown in parentheses 9 Expectednumber of deaths based on age-, sex-. race-. and cause-specific U.S. populationrates calculatedfor each five-year calendar period. 4 SMR is statistically significant at pc.05 if confidence interval (CI) does not include 100 286 I Mortality StudyIHanis et al .----- _--.....-.._-...- - Table 7-Standardized 8Nlortalil~:rlatio:; (!31F!;)fc.r lleaths Due tu Selected' Malicnant Neoplasms. 1970 Thmugh i -PI.... ..--._- _-- 1977 -Cancer Dea-th Site-!..._.-.-__- 0bserved/Expec!ed+ SMR Cl All malignant neoplasms 1.1 40-2091 666l705.0 94 87-102 Buccal cavity/Pharnyx (140-14!3 14A9.8 71 39-119 All digestive organs (l!iO-lSSi;i 196/195.3 100 87-115 Esophagus (150) 1Y17.2 87 49-144 Stomach (151) 33134.3 96 66-135 Large intestine (153) 67168.1 98 76-125 Rectum (154) 19121.1 90 54-140 Liver/gallbladder/bile ducts (1 55-156) 1911.3 133 74-219 Pancreas (157) 4Z39.2 107 77-145 All respiratory system (160-163) 217/231.6 94 82-107 Larynx (161) 719.9 71 29-147 Lung (162) 209R19.8 95 83-109 Eonelconnective tissue/skin/breast (170-174) 17114.3 119 69-190 Skin (172-173) 9a.1 99 45-188 All genitourinary organs (180-189) 100/120.6 83 67-101 Prostate (185) 56173.3 76 58-100 Bladder (188) 19125.3 75 45-117 Kidney (189) 2315.4 143 . 90-217 E!ain/CNS (191-192) 1 5113.0 115 65-191 All lymphopoietic (200-209) 55161.8 89 67-116 Lymphosarcorna/reticulosarcoma (200) 13111.4 114 61-196 m a (204-207l 20R5.9 77 q Other lymphatic tissue (202.203.208) 1st18.6 81 45-133 SMRs were not included when the observed number of deaths was less than five t ICDA. eighth revision codes shown in parentheses $ Expected number of deaths based on age-. sex-, race-, and cause-specific U.S. population rates calculated for each fne-year calendar period 5 SMR is statistically significant at p<.05 if confidence interval (CI)does not4nctude 100 pancreas (SMR =107), bondconnective tissuekkinhreast (SMR=119), kidney (SMR= 1431, and brain/CNS (SMR =115) and for lymphosarcomaslreticulosarcomas (SMR = 114) (Table 7). Among employees who were actively employed at some time during the study period, the observed numbers of deaths for several causes were significantly lower than expected. The SMR for all causes of death combined was 62 and the SMR for all malignant neoplasms was 76. How- ever, four of the six cancer sites for which slight excesses of deaths were observed in the total study population showed similarly elwated SMRs in this group. These were cancers of the liver/gallbladder/bile ducts (SMR= 139). kidney (SMR =1481, and brainlCNS (SMR =735) and lymphosarcomaslreticulosarcomas (SMR =130)(Table 8). To examine the possible effect of latency, employees who had had 15 or more years of employment (hired prior to 1956) at these plants were examined separately. The results of the comparison of the mortality experience of this group with that of the U.S. population were similar to those observed in the comparison of the total study population mortality with that of the U.S. population, except that the SMR o f 90 for all malignant neoplasms was significantly lower than expected (Table 9). The mortality ratios for white males were very similar to those of the total study population. Overall mortality ra- Journal of Occupational MedicineNol 27, No 4iApnl 1985 tios for black males were below 100 (all causes, SMR=97; all malignant neoplasms, SMR=92). There were a few causes of death for which SMRs were greater than 100, although none of these was statistically significant. They included diseases of the circulatory system (SMR=116), diseases of the genitourinary system (SMR-1201, and cancers ofthe digestive organs (SMR-131) (Table 10). -Cornparim of the Mortality Experienceof cohorts at Thm Separate Sites.With That of the US. Population Mortality among each of the three plant site populations was com- pared with that of the U S . population. Tables 11 and 12 provide extracts from these results. SMRs for major causes of death were included in Table 11 i f the SMR was statistically significant @<.OS) for any one of the three plant populations. For many causes of death, the numbers of observed deaths were lower than expected and many of the low SMRs were statistically significant. For the Bayway/ -Bayonne population, SMRs were high in two categories malignant neoplasms and endocrindnutritionaknetabolic diseases, (both of which were statistically significant). In general, the mortality from malignant neoplasms in Baton Rouge and Baytown compared favorablywith that in the US. population. SMRs for cancers of the brain/CNS and the kidney were 100 or above in all locations. None of these differences were statistically significant. In the Bayway/Bayonne population, elevations of the SMRs for 287 -Table 8 Slandardlted Morlallty Ratlos (SMRs) lor Selected' Causes 01 Death In Actlua Employeest 1970 through 1977$ Cause of Oeathg Observed Expected11 SMR All causes of death (000-999) All malignant neoplasms (140-209) All digestive organs (150-159) Liver/gallbladder/bile ducts (155-156) Pancreas (157) ' All respiratory system (160-163) y Lung (162) All genitourinary organs (180-189) Prostate (185) Kidney (189) Brain CNS (191-192) All lymphopoietic (200-209) Lymphosarcoma/reticulosarcoma (200) Leukemia (204-207) Diseases of Circulatory system (390458) Diseases of. respiratory system (460-519) Oiseases of digestive system (520-577) Oiseases of genitourinary system (580-629) 67011074.1 1831240.8 4ai6o.a 33.6 11113.1 65193.5 64188.7 23125.5 8l10.5 916.1 10i7.4 19/21- 8 U4.6 m.0 3401510.1 W59.3 32158.3 W10.6 62n 7611 79 139 84 70F 729 90 76 148 135 87 130 75 679 278 s.1[ 75 SMRs were not included when the obsenred number of deaths was less than five t Employeesactively employed between Jan. 1. 1970, and Dec. 31, 1977 $ Totai: person-years = 98.149; number of employees = 15,437; number of deaths = 670 5 ICDA. eighth revision codes shown in parentheses 1 Expected number of deaths based on age-. sex-. race-, and causespecific U.S. population rates calculated for each five-year calendar period 9 SMR is statistically significant at pe.05 all malignant neoplasms and all digestive organ cancers were statistically significant; excesses of respiratory system cancers were of borderline significance. Other cancers with greater than expected numbers of deaths (not statistically significant) in the Bayway/Bayonne population induded those of the buccal cavity/pharynx, braidCNS, and 'otherlsecondarylunspecifiedsites. Discussion Employees with less than one month of employment and nonregular employees were excluded from the study population. Because of their short .duration of employment, it is very unlikely that deaths among these employees would b e related to their employment in these petroleum and petrochemical plants. . Study follow-up was 98.7% complete, and it i s highly unlikely that the absence of information on these employees could have biased mortality results.' The 3,198 deaths observed during the study pe6od were distributed evenly throughout eight years, and official death certificates were obtained for 99% of all deaths. Thus, i t is very unlikely that more complete death verification would change the results. Data on variables of interest did not represent a problem in this study because the amount of missing data was low; work histories were available for 97% of all employees, and date of employment was available for 99.9% of all employees, and date of birth, race, and sex were known for all employees. In general, this population compared favorably with the 288 ' U.S. population for major causes of death. Some slightly elevated SMRs could be of biological importance even though they were not statistically significant. Employed populations have been estimated to have an overall mortality risk lower (on the order of 10% to 40% less) than that of the general population because of the so-called "healthyGmrker effect.'- The mortality ratio for all causes of death in this study was 91. In this investigation, the inclusion of a large group of retirees diminishes the impact of the healthy worker effect usually observed in industrial cohort studies. Most occupational epidemiology studies have not induded this group. The greatest proportion of this study population was employed before the beginning date of observation. When the workers who were actively employed during the study p e riod were examined separately, mortality amongthe active employees was much lower than that of the U.S. population. Infact, the SMR for ail causes of death after exdusion of the pre-1970 retirees was 62 as compared with an SMR of 91 when these retirees were included. Four of the six cancer sites for which greater than expected numbers of deaths were observed in the total population analysis were also observed to have SMRs greater than 100 in the actively employed group. These findings support the contention that any healthy worker effect probably OCCVK in younger populations with fewer years of employment and may not act on site-specific cancer mortality rates.' The possibilitythat the younger, short-termworkers could have diluted any occupational latency effect in the total study population was examined by analyzing the popula- Mortality StudylHanis et al -Table 9 Standardized Mortality Ratios (SMRs: for Selected' Causes 01 Death in Employees Hired , Before 1956,1970 Through 1977t I-. Cancer Death Sile+ Hired Prior to 1956 0bserved/Expectedg SMR All causes of death (000-999) All malignant neoplasms (140-209) Buccal cavitylpharynx(140-149) ,All digestive organs (150-159) Esophagus (150) Stomach (151) Large intestine (153) . Rectum (154) Liverlgallbladderhileducts (155-156) Pancreas (157) All respiratorysystem (160-163) Lung (162) All genitourinary organs (180-189) Prostate(1 85) Bladder (188) Kidney (189) BraiWCNS (191-192) All lymphopoietic (200-209) Lymphosarcoma/reticulosarcoma(200) 3.147/3,403.1 6131684.2 14119.3 181/190.5 1316.8 30133.5 63166.5 18120.6 14111.0 3rn8.3 201I226.O 193214.5 951117.5 51l72.1 19Q4.9 22I15.0 14112.0 44158.5 9110.8 9211 9011 73 95 77 90 95 87 127 99 89 90 81 7111 76 147 117 75 83 Leukemia (204-207) Diseases of Circulatory system (390-458) Diseases of respiratory system (460-519) Oiseases of digestive system (520-577) Oiseases of genitourinary system (580-629) 17124.5 1,94011,958.7 1621250.8 911125.0 51149.6 69 99 6511 791 103 ~~ SMRs were not included when the observed number of deaths was less than five t Total: person-years = 95,436.1;number of employees = 13.428;number of deaths = 613 f ICDA. eighth revision codes shown in parentheses 9 Expectednumber of deaths based on age-, sex-, race-, and cause-specific U.S. population rates calculated for each fwe-year calendar period U SMR is statistically significant at pc.05 tion of all workers hiredbefore 1956. The results indicated ently under way at Louisiana State University, whose pur- no dilution effect. The distribution of deaths and the short pose is to examine the high incidence of pancreaticcancers observation period precluded detailed analyses by la- in Louisiana. The excess of brain/CNS cancers observed tency. This problem can be solved in the future by in- in the Baytown population was not statistically significant creasing the observation period retrospectively and/or and not apparently associated with potentialfor exposures prospectively. An extended observation period would also on specific jobs. Further insight may be gained from a Gulf result in an increased number of observed deaths, which would improve the statistical power for examining more Coast case-control study to be conducted by investigators at the University of Texas and Louisiana State University.' detailed stratified analyses. In light of the publicity alluding to New Jerseyas "can- In comparison with the U.S. black population, black cer alley," digestive organ and respiratory site cancerswere employees were observed to have an excess number of examined using New Jersey and local county rates as the digestive cancers, mainly those of the stomach and pan- standard CTable 13). The differences between the ob- creas, but these were based on'very small numbers. These served and expected numbers of deaths for digestive or- diseases have been associated in other studies with socio- gan and respiratory site cancers diminished when these economic status, diet, and exposure to asbestos.. There rates were applied, suggesting that regional factors were was no way to examine these factors in this study because influencing mortality in this population. More detailed the data were unavailable. However, because this subgroup analyses using local and state rates as the standard for was very small, further analyses and speculations would each of the three sites would provide more information be unreliable. relative to geographical differences. , Comparisons of the mortality experience of employees Four other cohort mortality studies of petroleum and at individual plant sites with that of the U.S. population petrochemical workers whose results were published or indicated some areas for further study. The excess of pan- reported in the 1973 through 1982 period showed similar- creatic cancer deaths observed in the Baton Rouge plant population i s being reviewed in a case-control study, pres- ly low SMRs for all causes of death and for all cancer deaths,'*'' even though the observation periods and co- Journal of Occupational MedicineNol. 27. No. 41Apnl 1985 289 . - -Table 10 Standardized Mortality Ratios (SMRs) lor Selected' Causes of Death by Race, 1970 Through 1977 Cause of Death+ All causes of death (000-999) Malignant neoplasms (140-209) All digestive organs (150-159) Stomach (151) Large intestine (153) Pancreas (157) All respiratory system (160-163) Lung4162) All genitourinary organs (180-189) Prostate (185) Endocrine/nutritionaVmetabolicdiseases (240-279) Diseases of circulatory system (390-458) Diseases of respiratory system (460-519) Oiseasesof digestive system (520-577) Oiseases of genitourinary system (580-629) All external causes (E800-999) Total No. 01 person-years Total No. of employees Total No. of deaths White Malest -Obsened/Expec!ed 2,874l3.172.7 6031634.1 1711174.5 26i29.5 61162.4 35135.4 201R12.5 193l201.8 941103.6 51163.7 52/55.9 75311,809.8 150R35.0 831117.7 42f42.4 911166.6 SMA 91n 95 98 88 98 99 95 96 91 80 93 97 wi 7111 99 551 120,053.0 18.402 2,874 Black Males ObsenedExpectedg 284l293.7 54158.7 23117.6 714.4 514.3 613.3 14117.6 14116.6 6112.1 39.6 616.4 168t145.1 14117.1 7110.7 9J7.5 19R7.8 11.550.9 2.026 284 SMR 97 9.2~ 131 159 116 182 80 84 50 52 94 116 82 65 120 68 SMRs were not included when the observed number of deaths was less than five in either group +t Includes minorities other than black ICDA. eighth revision codes shown in parentheses 5 Expectednumber of deaths based OR age-. sex-. race-, and cause-specific U.S. populationrates calculated for each fweyear calendar period ISMR is statistically significant at p<.O5. -Table 11 Standardized Mortality Ratios (SMRs) for Selected' Major Causes of Oeath for the Three Plant Populations at Risk, 1970 Through 1977 Causa of Oeath Baton Rouge Baytown ObseNedl ObseNedl Expectedt SMR Cl$ Expectedt SMR CIS BaywayiBayonne Obsenredl Expectedt SMR CIS Al causes of death (000-999) Malignant neoplasms (140-209) EndocrineInutritionaUmeta- bolic diseases (240-279) Mental disorders (290315) Diseases of arculatory system (390-458) Oiseases of respiratory system (460-519) Oiseases of digestive system (520-577) Symptorns/senility/ilIdefined conditions (780-796) All external causes (E800-999) 1,205/1.320.8 249l272.0 18l24.6 116.2 729m.9 6W92.5 34150.0 8/19.5 52/81.o 91 86-97 784B82.5 92 81-104 1581209.9 73 43-116 9117.7 16 0.002-90 2f4.7 100 . 93-108 459534.4 71 55-91 48l68.7 68 47-95. 25140.8 41 18-81 17112.1 64 48-84 38/61.1 80 74-86 1.20911,213.1 ,75 64-88 2591223.3 51 23-97 32R1.6 100 94-105 116 102-131 148 101-209 42. 5-152 85 78-93 34.0 75 15-220 762f721.1 106 98-113 70 52-93 5w93.3 54 40-71 61 40-90 34139.8 86 59-120 141 82-226 17112.6 135 79-216 62 44-85 21155.1 38 24-58 Total No. of person-years Total No. of employees Total No. ot deaths Total No. Of cancer deaths 56,012.0 8.662 1,205 249 46,192.6 7,322 784 158 35,525.8 5,776 1,209 259 SMRs were included if SMR was statistically significant @<.OS) for any of the three plant populations . t Expected numbers of deaths were based on age-, sex-. race-, and cause-specific U.S. population rates calculated for each five-year calendar period $- SMR is statistically significant at pc.05 when the confidence interval (Cl)does not include 100 5 An employee may have been at more than one plant during this observation period 290 , Mortality StudyIHanis et al - ~ ~~~~~ ~~~ ~-~ ~ Table 12 Standardized Mortality Ratios (SMRs) for Oeaths From Major Categories of Malignant Neoplasms and Seleded Sites* for the Three Plant Populations at Risk, 1970 Through 1977 Baton Rouge Baytown BaywaylBayonne Cancer Oeath SHet ObservedlExpected$ SMR CIS ObservedlExpected# SMR CIS ObseNedlExpected~SMR CIS All malignant neoplasms (140-209) Buccal cavitylpharynx (140-149) P All digestive organs (150-159) Stomach (151) Large intestine (153) Pancreas (157) All respiratory system (160-163) Lung (162) Bonekonnective tissue/ Skinibreast (170-174) All genitourinary organs (180-189) Kidney (189) BrainICNS (191-192) All lymphopoietic (200-209) 2491272.0 m.7 92 81-104 26 3-94 7w5.3 97 76-122 11113.8 24iZ5.3 23115.2 79190.4 80 40-143 95 61-141 152 96-228 87 69-109 78/85.? 415.2 91 72-114 77 21-197 39146.0 85 60-116 915.8 155 71-294 515.0 100 32-234 25R3.2 108 70-159 158.209.9 75 64-88 516.3 79 26-185 39/56.4 69 49-95 619.4 11119.6 10111.7 54174.3 64 23-138 56 28-101 85 41-157 73 55-95 54l70.6 77 58-100 714.5 156 63-322 20130.9 65 40-100 614.9 123 45-268 614.6 131 48-286 16f18.4 87 50-141 259t223.3. 116 102-131 715.8 121 49-249 84163.7 132 105-163 1w11.1 32123.3 912.4 84167.0 144 82-234 137 94-194 73 33-138 125 100-155 77163.6 121 96-151 414.6 87 24-223 41143.7 94 67-128 714.7 149 60-307 4l3.4 117 31-299 14/20.1 70 38-117 If the SMR for a specific cancer site was greater than 100 for any one of the three plants, it was included t ICDA. eighth revision codes shown in parentheses $ Expectedvalues were based on age-. sex-, race-, cause-specific U.S. population rates by calendar year 0 SMR is statistically significant at pc.05 when the confidence interval (Cl)does not include 100 hort definitions varied considerably. Mortality due to lung and brain/CNS cancers and to leukemia, which appeared slightly elevated in two other studies,'L1s-lL was not found to be consisrently high in the present study or inthe other four large population-based petroleum/petrochemical cohort studies.'*1tl4 The differences in results may have resulted from the following: (1)one of the studiesls.u that showed elevated cancer mortality results was based on a proportional mortality ratio analysis, which tends to exaggerate proportional differences particularly for cancer mortality, and (2) the other study" was based on very small numbers of deaths. Similarities were observed across most studies for cancers of the digestive system and the genitourinary tract (slightly elevated in some subgroups). None of these el- evations was consistently or significantly high enough to warrant conclusions that any particular exposure was responsible for the increased risk. However, the need for further study has been emphasized.l**u Cigarette smoking has been linked as a factor in these diseases; until further analyses of these cohorts are available, taking smoking and other life-style factors into consideration, these results merely allow for speculation. Summary A cohort of 21,698 US. refinery and chemical plant workers was observed for eight years to compare their mortality experience with that of the US. population. The mortality inthis studypopulation was generally lower than -Table 13 Standardized Mortality Ratios (SMRs) for Selected Neoplasms for White Males at Baywaynayonne Expected Numbers of Deaths Based on New Jersey and County Rates for 1975 I Cancer Death Site Malignant neoplasms All digestive organs Respiratory system i New Jersey Rates ObseNed/Expected 2591248.4 84/75.8 84179.8 SMR 104 111 105 Countv Rate; ObseruedExpected 259.r262.3 84/82:2 84i83.8 SMR 99 102 100 Journal of Occupational McdicincNol. 27. No. 4/Apnl 1985 I 291 . . .. ' that of the U.S. population of similar age, sex, race, and calentlar year. Comparisons of the mortality rxpcricncc of the three plant populations with that of the US. pop-ulation indicated some arcas for further evaluation. References 1. Hanis. NM, Holmes TM. Shallenberger LG. et al: Epidemiologic study of refinery and chemical plant workers. / Occup Med 24:2(f3212.1982. 2. Manual of the International Classification of Diseases. Injuries and Guses of Death, eighth revision- Geneva: World Health Organization, 1967. 3. Monson RR: Apalysis of relativesurvival and proportional.mortality. Computcrs BIomed Res 7:325-332. 1974. 4. MacMahon 8, Pugh TF: Epidemiology Principles and Methods. Boston: Little Brown L Co.. 1970. p 4. 5. Vital Statistics of the United States, Val. 11. Mortality, p1.A National Center for Health Statistics, 1975. 6. Hanis NH. Shallenberger LG. Oupont ffi: Occupational cohort mortality studies: Do the benefits derived from tracingthose lost-tofollow-up justify the costs?Presentedat a meetingof the International EpidemiologicalAssociation. Edinburgh, Scotland. Aug. P-28,1981. 7. McMichad AI: Standardized mortality ratios a d the 'healthy worker effect': Scratching beneath the surface. / Occup Med 18:164- 168. 1976. n. COIV1'. K q w t t o I :(:dw c v 1pitlcoilctlci~:A Suoctruly (11 (:UIIIW~ FInrafonrcme,at1i9o7n9.. International Agency for Research on Gnrer. L_y_ons 9. Texas-Louisiana study seeks 1 0 clarify link between braincancer, refinery work. Ocrup Safety Health R e p 12:816-817.1983. 10. Tabenhaw 1, Cooper R. Gaff~.Wy : A Mortality Study of Fctro- leum Refinery Workers, report EA 7402. Washington. 0.C: American Petroleum Institute. 1974. 11. Hanis NM. Stavraky KM: Ten-yeu M d i t y study of employees in a petroleum company. Am I @idUniol106:224,19?7. 12. Hanis NM. Stavraky Kh(. Fowler IL: Cancer mortality in oil refinery workers: / Occup Med 21:164-174.1979. 13. Thcriault C.Coulr?tL: A mortality study of oil r d i n qwnr&crs. ' / OCCWhied 21367-370. 1979. 14. Rushton1,Alderson MR: An epidemiologicalsurvey of eight oil refineria in Britain. & / Ind M e d 36:Su5,1981. 15. Thomas TI., DeCoufleP. Moure-Eraso R: Mortality amongwork- ers employed in petroleum refining land petrochemidplants. / Oc- CUP Med 22:97-103,1980. n.16. Thomas Wucweiler RI, Moure-Enso R, et PI: Mortality p t - terns amongworkers inthree texas oil refineries. I Ocarp Med 24:135- 141.1982. Gentle Heroes If you are able, ...save for them a place inside of you and save one backward glance when you are leaving ..for the places they can no longer go. Be not ashamed to say you loved them, ..though you may or may not have always. Take what they have left and what they have taught you ...with their dying and keep it with your own and in that time when men decide and feel safe to call the war insane, take one moment to embrace ..those gentle heroes you left behind. --Untitled poem by Mpr Midud 0. O ' W l , reported missinginaction in.\netnam in March, 1970, published in"Vietnam letters: E c h o e s f m a War longCone" In US.Elcm and World Report. November 12.1984. 292 Mortality StudyManis et al