Document 0JKgrexNmm1yQoy9Om2nQ4bGR

Mortality from Leukemia and Other Diseases among Workers at a Petroleum Refinery C h a n s u d a Wongsrichanalai,MD,DrPH; Elizabeth Delzell,ScD; and Philip Cole,MD !Rtis study evaluates the mortality experience of 9484 white men who worked at a petroleum refinexy. We compared the numbers of deaths among these men during the period 1940 -- through 1984 with the numbem expected on the bas& of the mortality rates of US white men. Overall, there were 9874 obr&rrred comgved with 5668 expected deaths (standsrdized mortality ratio, 77). Mortality rates for most mw*orcause of death cstegvries and moat cancers were also lower than ex- pscted. However, there was a statistically signitrcant 50% exaem of leukemia deaths (44 obrved/89.6 expectad). Lym- paocstic leukemia was increcued both among men hired before lslo and among men hired in 1940 or later. In contrast, n&ocytic leukemia wna increased only among men hired in l W or later. The restriction of the myelocytic leukernis . ex- to men hired in 1940or later may be related t o p m e s s changes which occurred at the refiery &r 1940 and which d t e d in an increased percentage of benzene in certain rdhexy st-. me presence of an excess of lymphocytic bukemia, but not myelocytic leukemia, among men hired before 194osuogeetr that some &tor other than benzene was rcrponsiblefor the former condition. This study examines the overall and cause-specific mortality patterns of workers at a petroleum refinery in southern Illinois. The refinery started operating in 1918.Ita operationsnow include crude oil refining From the Department of Epidemiology, University of Alabama at Birmingham School of Public Health, Birmingham. AL. Dr WongsrichuuI8i in p m n t l y with the Department of Immunology, US Army lledical Component. Bangkok,Thailand. Addreas correspondence to: Department of Epidemiology, UAB School of Public Health. 209 Tidwell Hall, University Station, Birmingham,AL S5394 (Dr DelzeJI) A version of this report was presented at the 1988 American 0mUp.tion.l Health Conference in New Orleans. Louisiana, on April 98.1988. . arp9Ilss/es/sIos-olo.m/o Capyright 0 by American College of Occupational Medicine and production of gasoline, diesel fuel, aviation fuel, lubricating oils, and various petrochemicals. Both a follow-up study and a case-control study of leukemia among workers at the refinery have been done.'.' The follow-up study examined the cause-specific mortality experience of 3976 white men.' During the study period of 1973 through 1988,their mortality rates were lower than those of US white men for all causes, all cancers, and all other major disease categories. However, there was a twofold inc- in leukemia mortality, based on 14observed and 6.6expected deaths (standardized mortality ratio (SMR)= 213;96% con- fidence interval (CI)= 117 to 368). Eight of the 14men who died of leukemia had acute myelocytic leukemia (AML) (SMR = 394,95% CI = 172 to 788). Because benzene is present in petroleum rehneries and is gen- erally considered to be a cause of AML,' the observation of an excess of this disease was of particular interest and was evaluated further in a nested caae-control study? The latter investigation did not identify any specific job or department as responsible for the leuke- mia excess, and cornpasiron of the benzene expagure levels of cases and controls did not indicate any associ- ation d t h benzene. The present study further examines the mortality patterns of white men employed at the refinery. It is an ekpansion of the previous study in terms of both the number of workers included and the duration of the follow-up period. Methods The study includes all men not known to be black (they are referred to hereafter as "white men") who met the following criteria: (1) employed at the refinery for at least 6 months before Jan 1, 1984; (2)hourly and employed on Jan 1, 1942,or later; or (3) salaried and 106 Leukemia in PetroleumRefinery Worlters/Wongsrichanalai et al employed on J a n 1, 1940,or later. W e did not include hourly workers who terminated employment before 1942 o r salaried workers who terminated employment before 1940 because their employment records were not routinely available. Plant and corporate records provided the following information on each employee: name, Social Security number, company identification number, gender, race, birth date. dates of first employment and of termination at the refinery, employment status (active, retired, or terminated) and payroll classification (hourly, salaried, or both). Two mutually exclusive employee subgroups were defined on the basis of payroll classification. ``Hourly'' employees were men who had evpr been hourly, including those who had started as hourly and submquently tranoferred to aalaried peekions. "Salaried" employees were men who had never been hourly. Salaried employees were further classified as ever employed in production-related (PR) jobs or as employed only in nonproduction-related (NPR) jobs. A job was considered to be PR if i t involved at least 20% of an employee's time in areas with potential exposure to refinery chemicals. We used company, Social Security Administration (SSA) and Nationel Death Index records to determine the vital status of the cohort members as of Jan 1,1985. The SSA provided information on deaths occurring through 1984 and identified men presumed living as of 1988.Death CertifScateS were obtained from the compmy or from state vital records offices. A nosologist coded the underlying cause of death according to the InternationalClassification of We compared the mortality experienceof the refinery workers w i t h t h a t of the US general population, using '!the SMR as the measure of association. The S M R is bilefined as the ratio of the observed to the expected number of deaths, multiplied by 100.To obtain expected numbers of deaths, we multiplied the refinery workers' age- and calendar time-specific person-years of followup (grouped in 5-year categories) by the corresponding cause-specific mortality rates of US white men. For salaried workers, follow-up began 6 months after the first date of employment or on Jan 1, 1940,for men hired more than 6 months before 1940. For hourly workers, follow-up began 6 months after the first date of employment or on Jan 1, 1942,for men hired more than 6 months before 1943.Follow-up continued until death or until J a n 1, 1985. Men identified by the SSA as alive in 1985 and not reported as deceased in 1984 were assumed to have remained alive until December 31. 1984. For men lost to follow-up, person-years were accumulated up to the date of termination of employment at the refinery. We calculated approximate 95% CI of the SMRs assuming a Poisson distribution for the observed numbers of deaths. Results a r e designated as statistically significant when the two-sided 95% CI of the SMR excludes the null value of 100. A computet program developed by Monsonbwas used to calculate the S-. The program contained incorrect leukemia mortality rates for 1975 through 1979 and 1980 through 1984. We corrected the rates for 1975 through 1979 as suggested by Monson (personal communication to USDR users, May 1, 1987) and replaced the 1980through 1984 rates with rates calculated from published US vital statistics data for 1980.6 Expected numbers of cell type-specific leukemia deaths were derived from 1968 US mortality data7 for the period 1950 through 1969 and from Surveillance, Epidemiology, and End Results program mortality rates' for the period 1970 through 1984.General pop ulation mortality rates for specific leukemia cell types were not available before 1968.It is likely that cell typespecific rates increased during 1950 to 1970 88 a result of more accurate diagnosis and reporting and, poseibly, a true rise in incidence and mortality. If this were the case, the use of 1968 rates as referent rates for the period of 1980 through 1969 would overestimate the true expected numbers. Expected numbers of cell typespecific leukemia deaths were not estimated for the period 1940 through 1949.There was only one leukemia death in the study group during this period. Of the 9484 men included in the study, 7524 (79%) had been hourly at some time during their employment at the refinery, 1871 (W%w)ere always salaried, 47 were salaried, unknown if ever hourly, and 43 had an unknown payroll classification. Of those who were 81ways salaried, 1109 (59%) had held PR jobs. About 97% of the men were hired before 1940,34% in 1940to 1949, and 39% in 1950or later. Fiftysne percent had worked at the refinery for at least 10years as of Jan 1, 1984. A t the close of follow-up, 6265 (66%) men were assumed living, 2874 (30%) had died, and 345 (4%) were of unknown vital status. The total number of person-years of follow-up was 267,540,with an average length of follow-up of about 28 years. Death certificates were located for 2,733(95%) of the men who had died. The 141 decedents for whom a death certificate was not available were included in the S M R for all causes combined (see below). However, cause-specific observed numbers of deaths and the corresponding SMRs were not adjusted to include a percentage of deaths from unknown causes. Therefore, cause-specific SMRs are underestimated by an average of 5%. The SMR for all causes combined was 77,reflecting approximately a 20% deficit in overall mortality fer the refinery workers compared with US white men (Table 1). Observed numbers of deaths for most mejor disease categories were fewer than expected, and most of these deficits were statistically significant. Table 2 displays data for specific cancers. The observed number was approximately equal to or was leas than the expected number for all sites except leukemia (44/29.6).The SMR for leukemia was 149 (95% CI = 10s to 200). There were statistically significant deficits of deaths from buccal cavity and pharynx cancer (6/ 23.5, SMR = 26, 95% CI = 9 to 56) and from lung cancer (210/249.3,SMR = 84,95% CI = 73 to 96). Joumrlof OcmpdoW Medidne/Vdume 31 No. 2/Fehary 1989 107 1 Men with long duration of employment and those with long follow-up also had statistically significant deficits of deaths for most major disease categories. Employees who had worked at the refinery for at least 10 years had 8n SMR of 75 (95% CI = 71 to 78) for all causes ~ and an SMR of 82 (95% CI = 75 t o 91) for all cancer. M e n observed for at least 20 years since starting work had SMRs of 80 (95% CI = 77 to 84) and 87 (95% CI = 80 to 95) for all causes and all cancer, respectively. Table 3 displays result0 for selected causes of death separately for hourly and salaried workers. These two groups had generally similar mortality patterns. However, the relative excess of leukemia was higher among salaried (9/3.7, SMR = 242.95% GI = 110 to 459) than TABLE 1 Observed Numbers of Deaths and Standardized Mortality Ratios (SMRs) for Major Causes of Death Among Refinery Wwkers. 1940 to 1984 Cause of Death 0 b S W ~ d SMR 95%CI' All causes All cancer infectiveand parasitic diseases Benign neoplasms Oiabetes mellitus Diseasesof the Blood Nervous system Circulatory system Respiratory system Digestive system Genitourinary system Senility and illdefinedconditions Ai external causes Other known causes 2874t 661 14 8 48 9 14 1441 136 108 28 19 228 17 77 74-80 86 80-93 23 12-38 78 34-155 90 67-120 92 42-176 42 23-70 74 70-78 57 48-67 62 51-74 50 33-73 51 31-79 74 65-84 49 29-78 CI, confidence interval of the SMR. tlndudes 141 persons for whom a death certificate was not available. among hourly workers (35/25.7, SMR = 136,95% CI = 95 to 190). A direct comparison of the leukemia mortal- ity rates of salaried and hourly employees yielded a rate ratio of 1.9, with a 95% CI of 0.9 to 4.0. Within the subgroup of salaried employees, there were no striking differences between men in the PR and NPR jobs. The leukemia SMR was 192 (4/2.1) among men in PR jobs and 307 (5/1.6) among men in NPR jobs. In addition, hourly, but not salaried, workers had small increases in deaths from skin cancer (13/10.7, S M R = 121). Hadg- kin's disease (8/6.7, SMR = 120). and other lymphatic and hematopoietic tissue cancer (19/16.4, SMR = 116). Further examination of leukemia mortality patterns in the overall cohort indicated that the leukemia excess appeared to be largest in the 1950s (10/3.8, SMR = 263, 95% CI = 126 to 484),although smaller excesses also occurred in the 1960s (11/6.9, SMR = 160,95% CI = 80 to 287) and the 1970s (16/10.6, SMR = 151, 95% CI = 86 to 245). There was no excess in 1980 through 1984 (6/7.0). Only one leukemia death occurred in the 19409 (SMR = 76). There was no clear pattern of increasing leukemia SMRs with increasing duration of employment. However, the SMR was somewhat larger among men who had worked at the refinery for at least 10 years (33/ 19.8, SMR = 167, 95% CI = 115 to 235) than among men employed for fewer than 10 years (11/9.8, S M R = 112, 95% CI = 56 to 900). There was no excess of leukemia within 10 years of first hire (1/2.0). The largest excess occurred during the period 10 through 39 years since starting work (35/19.1, SMR = 183.95% CI = 127 to 254), and there was, again, no excess after 40 years since first hire had elapsed (8/8.5). Evaluation of leukemia mortality by period of hire and period of death revealed that, among men hired before 1940, there was a 2.5-fold increase in leukemia TABLE 2 Observed Numbersof Deaths and SMRs for Sekted CancersAmong Refinery Wwkers. 1940 to 1984 CmlcerSim obsmwd SMR 95% CI All sltes Buccalcavlty and pharynx Esophagus stomach Large LIVW intestine and- rectum Panmeas Larynx Lung Skin Prostate Bladder Kldf-U3y Centralnervous system All lymphatic and hematopoletic tissue' Lymphosarcoma Hodgkin's disease Leukemia Other lymphatc and hematopoletchssuet Other and unknown 661 6 13 27 92 ' 11 31 5 210 15 44 17 18 19 90 14 9 44 20 63 86 80-93 -26 9-56 72 38-1 23 67 44-98 97 78-119 67 34-120 75 51-106 44 14-103 84 73-96 118 66-195 79 58-1 07 71 42-114 95 57-151 89 54-139 126 101-155 99 54-166 114 52-216 149 108-200 105 64-162 89 67-114 'Three deaths attributed to myelofibrosisare included in the category of "all lymphatc and hematqmettc tissue" but not in any of the subcategories. t Includes other lymphoma (n = 141, multlpk myeloma (n = 5) and pdycythemla vera (n = 1). 108 Leukemia in Petroleum Refinery Worfters/Wongsrichanalai et el TABLE 3 Observed Numbers 01 Deaths and S M R s by Payroll Classikxtm Cause ol death All causes All cancer Buccalcavity and pharynx Esophagus stomach Large intestine and rectum Liver Panaeas Lung Skin Prostate Bladder -Kidney centnd nervoussystem LHr Lymphosarcoma Hodgkn'sdisease Leukemia 0therLH-l Otherandunknowncancer Circulatory diseases Respiratory diseases Extemal causes Other knowncauses Unknown W r f y (n = 7524) Obtaned SMR 95% CI 2589 585 5 13 22 84 10 26 189 13 41 17 16 14 -78 13 8 35 19 57 1316 129 199 238 1.-2_2 79 76-82 87 80-95 24 8-57 83 44-141 62 39-94 101 80-1 25 69 33-127 72 47-105 88 76-101 121 64-207 82 59-112 80 47-128 98 56-159 78 42-130 127 100-158 106 56-182 120 52-237 136 95-1 90 116 70-181 80 61-104 77 73-81 61 51-73 78 66-90 57 50-65 Smlwhd (n = 1871) 0b.arVed SMR 45% CI 2~63~ . 68 1 5_ 9_ 52-66 72 56-92 34 0-191 0 0 0-167 4 94 25-241 8 74 32-1 46 1 55 0-304 4 81 22-207 19 59 36-92 2 107 12-386 3 -56 11-162 0 0 0-1 49 2 83 9-301 4 128 35-329 12 129 66-225 1 54 0-302 1 84 1 - 4 6 9 242 110-459 1 40 0-224 8 76 33-151 117 54 45-65 6 23 8-50 26 51 33-74 29 48 32-69 17 LHT, lymphatic and hematopoietic tissue. (17/6.8,SMR = 251.95% CI = 146 to 402)during 1940 through 1989 and no excess thereafter (7/8.1).In contrast, among men hired in 1940 o r later, no excess of leukemia was present before 1970 (5/5.2),whereas a 67% ex- (15/9.6,SMR = 157,95% CI = 88 to 859) occurred after 1989. '11 Table 4 displays cell type-specific observed and expected numbers of leukemia deaths by decade of death. The excess of lymphocytic leukemia was highest in 1950 through 1959 (5/1.0,SMR = 505, 95% CI = 164 to 1180), declined in 1960 through 1969 (6/2.1,SMR = 282,95% CI = 109 to 615), and was absent in 1970 through 1984(5/4.9).There was no excess of myelocytic leukemia in 1950 through 1959 (0/1.8) or in 1960 through 1969 (2lS.O).However, in 1970 through 1984, there was a greater than twofold increase in deaths from AML (1!2/5.5,SMR = 219,95% CI = 11s to 583). Monocytic leukemia was increased from 19M)through 1979 (6/1.1,SMR = 570,95% CI = 809 to 1241), but there was no death (0.27 expected) from this disease after 1979. Analysis of cell type-specific leukemia by period of hire also revealed different patterns for lymphocytic and myelocytic leukemia. Men hiired in or after 1940 had a 1.6-fold increase in deaths from myelocytic leukemia (1 1/6.9).whereas there was no excess of deaths from this disease among men hired before 1940 (4/6.1). In contrtret, the excesses of lymphocytic and monocytic leukemias were not related to period of hire (lympho- cytic leukemia among men hired before 1940,9/4.5; among men hired in or af'ter 1940, 7/9.6)(monocytic leukemia in men hired before 1940,4/0.67; among men hired in or after 1940,%/0.65). TABLE 4 observed Number of Leukemia Deaths snd SMRs' Accorcbng to Decaeeof DeathandWType Decade 1950-1959 1960-1969 1970-1984 Cell Type AHlymphocybc All mvekcvtic Mondcybc' Other and unspecitied All lymphocyt~ All myebcyhc MonocytE OtWandUnSpeahed Lyt~~phocyt~ Acute Chronic other MY-* Acute Chronic Other Mvc OtherendUnSpecmed ob..mrdsMR 5 505 0 (1.8)t 1 435 4 531 6 282 2 67 3 731 0 (1.4) 5 102 1 130 4 119 0 (0.76) 13 158 12 219 1 44 0 (0.48) 2 292 2 55 Expected numbers for computing SMRs were estimated from US V i Statistics, 1968' for deaths in 1950 through 1969 and from other data*for deaths h 1970thr- 1984. The 1968 data do not distinguish betwesn acute end ciwonic fams of 'and myelocyte leukemia. t The expected number is providedwhen the observednumber is zero. Six men with leukemia mentioned on the death certificate but not coded as the underlying cause of death were not included in any of the above analyses. Three of these men were hired before 1940;two of them had lymphocytic leukemia, and one had aubleukemic leukemia following myelofibrosis. The other three were hired --fo /Volume 91 No. 2/February 1989 . 109 after 1940; two of them had unspecified leukemia and one had AML following chronic myelofibrosis. The 9484 workers included in this study had an average of 28 years of follow-up during the 45-year period of 1940 to 1985. Their mortality rates for all c a w combined and for most major cause of death categories were lower than thoee of the general US population. The overall favorable mortality experience persisted among long-term employees, and it was still present after a0 years of follow-up. These results are consistent with most previous studies of petroleum workers." h n e r a l factors that may be responsible for the apparent health advantage of these workers include selection of relatively healthy workers for employment in the industry and maintenance of their healthier status through benefits deriving from the economic advantages of employment.20Moreover, these fndings point to the absence of occupational hazards within the industry that might have a pervasive, negative impact on employees' health. The other noteworthy results of this study pertain to patterns of mortality from various cancers. First, there were no excesses of certain cancers found to be increased in other groups of r e 5 e r y workers, including cancer of the brain, lung, stomach, pancreas, bladder, and kidney?' Among hourly employees, however, there were 8mall. statistically nonsignificant increases in deaths from skin cancer, Hodgkin's disease, and other lymphatic and hematopoietic tissue cancer. Studies of other p u p a of refinery workers have reported inconsistent msults for each of these diseases. 8eCond. there was a s t a t i a t i d y significant 50% ex- ''c888 of leukemia. Only three". la, of nine"' investiga- tions of other groups of refinery workers have reported leukemia excesses of 10% o r greater. Morgan and Wong" found a leukemia SMR of 1 7 3 (95%CZ = 109 to m),whereas the other two studies reported statisti- a l l y nonsignificant leukemia exceasea of less than So%. The overall leukemia excess found in the present otudy may be due to exposures at the refinery, to unidentified nonoccupational exposures, or to exposures at work sitea other than the refinery. Chance is another poseible, although unlikely, explanation of the observed leukemia patterns. Several observations argue to some extent against the hypothesis that the excess is due to refinery exposures. Salaried workers had a larger leubemia excess than did hourly workers, although the latter group probably experienced the higher exposures to chemicals at the refinery, and there was little difference in the magnitude of the leukemia excess between salaried workers in PR and NPR jobs. Also, there was no consistent increase in the leukemia SMR with duration of employment. However, these results do not detract greatly from a causal interpretation of the obmrved association because payroll classification and overall duration of employment may not be strongly correlated with leukemogenic exposures, if any, at the refmery. Evaluation of leukemia mortality by cell type revealed excesses of lymphocytic, myelocytic and monocytic leu- kemia, some of which were dependent on period of death and period of hire. Specifically, there was a marked shift from an excess of lymphocytic leukemia among deaths occurring before 1970 to an excess of AML in 1970 through 1984. Furthermore, the myelocytic leukemia excess was restricted to men hired in 1940 o r later and dying in or after 1970, whereas lymphocytic and mono- cytic leukemia was increased among men hired both before 3 9 4 0 and in or after 1940. Reasons for these patterns are not clear. Diagnostic techniques for cell type-specific leukemia have improved during the past few decades, and misclassification be- tween myelocytic and lymphocytic leukemias in 1950 through 1969 may explain some of the change over time in cell type distribution. Benzene, which is present along with other aromatic hydrocarbons in oil refineries, is considered a cause of AML3 but not of leukemia of other cell types. Studies of rubber workers exposed to benzene, toluene, and other organic solvents have suggested that these agents are associated with chronic lymphocytic l e ~ k e m i a . ~H. o*w~- ever, there is no firm evidence relating specific chemical agents to leukemia cell types other than that relating benzene to AML. Manufacturing processes at t h e refinery have changed over time, leading to an increase in the number of units that produce or use benzene and in the percent- age of benzene in refiiery streams. The concentration of benzene in refinery streams was not more than 5% until 1941, when toluene units were built with one stream of 5%to 10% benzene. Benzene was also suppiied by tank cars for the cumene plant (operating from 1943 to 1945) and the dodecyl benzene unit (1950 to 1954), and a coke oven product containing 65% to 88% benzene was delivered to the refinery from 1966 to early 1983. Benzene extraction units and catalytic reforming units, in operation since 1960,have benzene streams of 5% to 100%. However, it is important to note that the in- creased amounts and concentrations of benzene at the refinery were accompanied by enhanced efforts to con- trol exposure. Thus, the percentage of benzene in refin- ery streams may not have been correlated directly with actual exposure. & Assuming that the leukemia mortality patterns by cell t&, period of hire, and period of death are real and not due to disease misclaaeification, they have sev- eral implications: (1) The restriction of the myelocytic leukemia excess to men hired in o r after 1940 may be related to changes in exposures at the refinery in the 1940s o r later. One possible explanation is an increased exposure to benzene resulting from the process changes that occurred at the refinery after 1940. However, this explanation is spec- ulative because the changes may not have resulted in a concomitant increase in workers' benzene exposure lev- els and because other exposures, possibly relevant to leukemia, may have been introduced during this period. Also, the present study did not include a direct evalua- tion of the relation between leukemia and benzene or i 4 110 Leukemia in PetroleumRefinery Workers/Wongsrichanaleiet ai other chemical exposures, and an earlier case-control study of leukemia at the refinery found no association with benzene.* (2) The fact that an excess of lymphocytic leukemia was found among men hired before 1940 suggests that benzene WBB not entirely responsible for the overall excess of this cell type, because there was no corresponding increase in myelocytic leukemia among men hired 6. National Osnter for H d t h Statistics. ViW Statlatics of the United Stata, 1980.vol e, Mortality, Part A. DHHS Publication No. (PHS)85-1101. Wasain@on. DC. Department of Health and Hurmrn Services, 1985. 7. National Center for Health Statiaticn. Vital Statistics of the United States 1968, vol e, Mortality,Part A. Publication No. (ISM) 72-1101. Wmhington. DC,Department of Health, Education and Wel- fare, 1978. 8. National Cancer Institute. Surveillance, Epidemiology and Ehd Results: Incidence and Mortality Data 1915-1977, N E I Publication before 1940. NO.81-8890. B s t h d a , National h t i t u t e s Of H d ,1891. (8) Finally, the lymphocytic leukemia excess dimia- ished in 1970 through 1984.It is possible that this trend is re81 and reflects a reduction in some unknown leukemogenic exposure related to this cell type. In addition, 9. Rurhton L, Aldemn MR: An epidemiological survey of eight oil refineries in Britain. Br J Ind Med 1981;88:895-gSI. 10. Schottenfeld D,Wsnhauer ME.Zauber AQ,et al: A prcwpective mtudy of morbidity and mortality in petroleum industry employees in the United Ststsr--. pre)Uminaq report. In:Pet0 R, Bchneidermann the AML excesa was lower in 1980 through 1984 (3/ 8.1,SMR = 140) than in 1970 through 19$9 (9/3.3, ShIR = 271), and there have been no deaths from moncrcytic leukemia in the 19808. However, the expected numbem of cell type-specific leukemia deaths in 1980 M (ed.):0 ~ ~ t i d O . toff0oc~Cupatiod Ckoosr. Cold Spring Hubor. NY,Cold Spring Harbor Labomtory Prea, 1831.pp S47-WS. 11. Wen CP, T d SP,M a C l d l u ~WA, et d:hnp-term morWrty atudy of oil re5nerg workers: I. Mortality of hourly and nluid workers. Am J & i d d o 1 lsBB;118:588648. 18. Morgan RW, W o w 0 An epidemiologic d y s b of the mor- through 1984 were rather small, and future observation of leukemia mortality will be necessary to evaluate the apparent reduction in the leukemia excesses in this group of workers. M t y experienw of Mobil 011Corporation employeem at the Beaumont. Texan re5nery. Oakland, CA,Environmental Eealcrlth Amooiitea, Inc., 1984. 13. Hanb NM,Shallenberger M,Donaleaki DL. et al: A retapective morblity atudy of worker8 in three mqior US. reflneriea 4 chemical plants. Part 1: CompariaoM with U.S. population. J Oooup Med 1985;87:885-e98. 14. Divine BJ, Buron V, &plan SD: Te- mortality study I. Acknowledgments Mortality among refinery. petrochemical, and reseuch workers. J Occup Med 1985;87445447. Thin work waa supported by the Shell Oil Company,National Cancer Inrtitute Grant CAS9988, and an award from the Shell Oil Company 15. Wong 0, Morgan RW, W e y W J . et al: & epidemiological study of petroleum reilnery employees. Br J Ind Med 1986;+8:6-17. 16. Keplan S D Update of a mortality atudy of workers In petmleum Foundation. reheriea. J Occup Med 1986;88:614416. 17. N e h n NA. Van Peenen PFD,Blanchard AG: Mortality in a recant oil refinery cohort.J Oocup Mad 1897;893610-813 I References 18. Theriault 0,Provenchar 8. Mortality rtudy of oil rehorg workers: Five-year follow-up. J Occvp Med 1987;89:867-880. 19. Divine BJ, Barron V Tezaco mortality study: m. A cohort `h 1. M&w DS,Joyner RE,Cole P Ex- leukemia in a refinery population. J Occup Med 196&3785XM&3. 2. Austin H, Cole P.McCraw D8:A cacw-control study of leukemia at an oil refinery.J Occup Med 1986;8831169-1179. study of producing and pipeline workers. Am J h d Med 1987;11:188- m. 90. McMichael AJ: Standardizedmortality ratior,and the "healthy worker effect": Scratching beneath the eurisce. J Occup Med 3. Uac Monographs on the Evaluation of the Cawinogenic Risk 1976;18:166-168. Of Chemicalr to Man:Some industrial chemicals and dyest-. Lyon, 81. Savitz DA, Moure R. Cancer risk among oil refinery worken: France, lARCMoD0gr lSge;89:98-148. A review of epidemiologic atudien. J Occup Med 1984;86:66&?-670. 4. National Center for Health Statistics: International CLMSiflcation of Disease. adapted for use in the United States,vol 1,Sth Rev. 22. Checkoway H, Wilcosky T, Wolf P, et al: An evaluation of the associationsof leukemia and rubber industry solvent expos-. A m J D m W Publication No. (PHS)719. Washington, DC,Department of 2nd Med 19&9,6:e89-&49. Health. Education. and Welfare,1969. e8. McMichael AJ,Spirtas R, Kupper U:An epidemiologic study 5. Monson RR: Analysis of relative survival and proportional mortality. Cornput Biomed Res 1974;7:336-833. of mortality within a cohort of rubber workers, 1964-7S. J Occup Mad 1984;16:458-464. . occupatknrl Med&ine/Vokrmo 31 2/Febrwy 1989 111