Document 70M75ye5jEVY3yxMjvVaBY95o

-.. generosity 4 scientists rigin in the itional and ikoff, MD w-in-chief abomtory, Medicine, ;;rY10029 Am& J o u d of Industiial:-Mcdiie63-16 (1984). .- -,-,+ ' -* A '.- -.e;:.. I ,5.1. . Cancer Mortality Patterns by Work Category in Three Texas Oil Refineries Terry L. Thomas, MS, Richard J. Waxweiler, pho, Michael S. Crandall, MS, Deborah W. White, BS, Rafael Moure-Eraso, PM), and Joseph F. Fraumenf, Jr., MO ofcases and anratEhedset of controls whodicd from other causcsof death werc obtained from company personnel records. Work historia were summaze$ by classifyingeach job tide and depanmem envy into one of several broad work' catqyjricqo&&nery:Unit operations. A worker was considered "exposed" to a work category if he was known to have worked at least 1 day in the category IS or more years prior to hisdeath. Maximum *likelihood esCimatcs of the rrlacive risk for brain tumor. stomach cancer. and leukemia wcrc calculated by work category using a procedure for matched1- data. No strong assochtions for brain tumor risk were scen with any work categories. A sli ht wociatjon for leukemia was secn among W&IS in the Treatin catego kilmw Unlt operauonr that rcdtlce the I m l or ammatIc and s u k r cms&nis of penolam producu and combine them with additives to imprwe their quality. St& cancerrisk was dcvatedamong maincernur~workers and w o h exposed to lubricating OiL and ppraffin wax processing. INTRODUCTION ... 3- ~ 3 A Proportionate mortality study of 2,132 deceascd m q b e r s of the Oil, Chemi- ~ a l .and h m i c w o k ~nternationaUl nion (OCAW) &~oyed in.& ~ e x oail~ refineries indicated elevated proportionate mortality ~ & L P M R S L M &N~~ O K . q e a and leukemia among white males (TablzfF)omas ct al, 19821. Excess mortality from6-st occurredamo?npyhiA,$on members. '-I I c .- 4 Thommetal PMRS were elevated for other hematopoietic and lymphatic malignanciesand cancers of the skin, prostate, and pancreas. These results suggested the possibility of occupational risk faders. The prcrent report is an d u a t i o a of work histotiesof deeased active and retired union members who had worked at these refineries to determine whether deaths from brain tunor, stomach cancer,leukemia, or ocher selected 11~0plasm d with greater fresuency thanapcctcd aaong unployecs in particular work categories. MATERIALS AN0 METHODS Study Subjects T h i i - s i x deaths fram brain tumor, 48 from stomach cancer, and 33 fmm leukemia w e n observed h c c n 1947 and 1977 in the p n ~ p o t t i om~or~ta~lit~y study of active and retiraideceascd OCAW memberswhohadworlcedatthethrrenfmrics momas ct ai, 19821. One brain tumor case, four stomach cancer cases, and four leukemia cases subsequently identified from records of active and ruircd daceased OCAW members w a e added to the originaldata for the present study. One leukemia case was eiiminatui from the study when additionalinformation indicated that we had not obtained the comct death certificate. Because the brain is a commonsite of muasmcs, haspitdrecordsand pahlogy reports w e n sought for each brain Nmor case to comborate dcath certificate diagnoses. Alttrough many ofthe deadu occumdmocethan 15yeamap, hospital rccords w a e obtained for 28 of the 37 brain Nmoc cases.Siiof the 37olscs wen excluded fmm further analyses because ~ I Csuppomng diagnostic information available indicatcd that no brain tumor was present or that the brain was nor the primary site. For each d e with brain tumor, stomach cancer, or ladcunii three dcaascd controls w e n selected h t h c active and retiredunion members whohad worked at the same rcfineiy,w m of thesame nux and swt,anddiedofcauses other than those bcing investigated. The sciectcd conaois were those who matched as c l d y as possible on age at death, date of death, aad date of tirst wnbershl'p in the union. Complete work histories for cach casc aad control were rcqucsted from the three companies. Records w e n not located f o r m leulcemiacascs; thus, these cases and d r e i r m a t c h e d c o n t r o i s w e n ~from dre study. Records for 16 conaols weft not locatcd, neccssifalingselectionof substituteCMlttOls. Records fbr eight substitute controls could not k obtained and they wuc diminated from the SaIdy. 'chilty*ne and cancers ' ofoccupaof denascd * determine +ctcd nen*nparticular -d 33 from d i t y study *erefineries . s. and four .d dcaascd ;relcukania chat we had -1 pathology 'ticatediagical rccords ;% excluded itable indi:site. 'X& .orkcd at thanthose cIoscly as the union. I the three .' cases and tltrols w e n If substinne Thirtyane i .. CmxrMortnlitybyWorlrCntcgory 5 brain tumor, 52 stomach cancer, and 34 leukemia oiscs and their matched controls wcrc included in the finalanalyses. Work histories w e n also obtained for cases of multiple myeloma (N = 9). Hodgkin's discase (N = 9), non-Hodgkin's lymphoma (N = 2% and cancers of the skin (N = 14). prostate (N = 51), and p c r c a s (N = 41). These GISCS were ascertained using identical procedures as dtscribed above. Individually matched controls were not selected for these cases because the pool of eligible controls was deplatd. Instead,the controls chosen for brain tumor, stomach cancer, and leukemia were pooled and used to develop crudc cstimarcsof usual employment pattuns within the refinery work categories. The work histories of thcse cascs were reviewed for cvidcnce of any unusual distribution by work category. WO& category aassificatlo~ The three refineriesare located in the &aumont/PonArthur area of the TCXS Gulf coast and have been in operation since the discovay of oil in that region approximattiy 80 years ago. Although they am owned by differtnt companies, oprrations at the refineries are very similar and include standard proctsses uscd to convert crude oil into fuels, motor oils. and various petrochemicalproduas. Work histories for each subject wen summanzed by categorizing the types of unit operations common to each of the refinaits. The identification of the unit operations by categories was done according to the American Petroleum institute (MI)description of an "average" refincry [MI,19731. The following mutually exclusive categories were created, and the rcfinay pmcesc~and unit operations included in &are shown in Figure 1. category 1 (crude oil). This category includes the "first-step" refining pro- cessesof distillation and cracking. Distillation separates crude oil into s c v d basic strcams: (a) wet gas and gasolincr, (b) light naptha; (c) kerosene, gas oil, and middle distillates; (a) heavy gas distillates; and (e) red& crude. Strcam (a) is used to product g a s o i i Stream (b) is the raw material used by the catalytic Reforming Unit to produce gasoline as well as middle distillates used by processes in Category 3. stream (c) is the raw mattrial uscd by one of the processes in Category 3. Stream (d) is sent to the Catalytic Cracking Unit which products "cracked" gasoline used in Category 3. The Vacuum Distillation Unit piocesses stream (e) to product heavy gar oil used by the Catalytic Cracking Unit, lube stocks processed in Category 2, and solid residuum for Category 4. a.category 2 (lube This cattgory indudes those "SccOnd-stqJ" reflniig proctsses n e a s a t~o~prodwx lubricating oiis, paraffin wax, and raw materials for catcgory5. pataftinwaxesarescparatedfromlubricatingoiis in the PropaneDewaxing Unit and in the SoIvent Dewaxing Unit which uses phenol. benztne, methylahyl ketone. or toluene. Fg- processes include wax pressing and the blending of additives with lubricating oils after dewaxing. categorg3 (treating). Gasoline, kcmscnc, light fuel oils, and d i d oils are produced by units dassificd into this category. PFocesscs inciude blading of addi- tives, hydrog.en t.reat@ and Polvmatzlloon to to rcmove sufur, alkylation to producehigh-oaane imprwetbequality ofgasolirreproducu. products, Category 4 (cdring). Solid residuum from dw:Vacuum Distillation Unit and other =fining processes are uscd by units classified in this category to product parolaml coke. CaegOrY 5 (grease Piant). In this category, lubricatingoils arc combined with soaps, detergents, and mineral additives to yicld various indusvial lubri~anaand grtases- - - -------- -- ca duties, c;, Altl large-sizc occurred in Figure Eac intoone substance. 00 classif! undassifi.. OpCratiOfk. difficult f(. The aging grrase pac classified included :: *tion a, arcaS wen cations wc plant at R, k.1. arbitrarily atleastlct the latenq of an -ex[. Thetotal. -exposed- quircmel. categories. Statistical Maxi, dence inter -=w variable m. `cfincry WC!. identi. woI1ccd in tbase Who \? ifthe9096 usedisnot. dative imp. Trrrrdh Procedun 1: grratu, the =mrols. anL tTr- _-__ LI .- - ---_ - 1-m nbined with Iricants and ses for the ishing. aintcnance, ed into this ;ory, which tegories of 'ry, carpen{ard labor, wide more 1% andyard =.-, mping and lity control nyinclude the R ref- CIUmr Mortality by Work Category 7 Category 11 (other activities). This category includes office work, guard duties, cafeteria work, and other presumably "lowcxposure" activities. Although the unit operations descn'bed above represent the basic units of a large-size refinery (more than IoO.OO0 barreidday), many process changes have occurred within the refining industry over time. Some of thcse changes arc designated inFigurr1. Ea& job title and department envy in a study subject's work history was placed into one of these categories, but no qdtative exposure estimate to particular substancts was attempted. If a department or unit was not listed, an atttmp was made to classify the entry solely M the basis of job atk. About 6% of the entries w e n undassifdle and w e n treated scparatdy as 'unknown." The complexity of refmery optrations and diffcnnces in departmental groupings between the companies made it difficult to establish a consisccnt category dassificationacross the three plants. The following inconsisccncies w e n not rrsolvak First, refinery A had a pack- aging and shipping operation that was separate from the grease plant: therefore, grease packaging operations were not included in Category 5 for Refinery A but were classified into a separate category. Second the Receipt and Movement category indudcd all pump houses at Refinery C. However. owing to the departmental orgad o n at refineries A and B. jobs in pump houses in the Crude Oil and Lube Oil . artas w e n cIassified into Categories I and 2, rrspcctively. Finally, separate classifi- cations w m made for the chemical divisions a! Refincries A and B. the Compnssion Plant at Refinery A, and the Alcfilor Unit, a watcr-rrcating unit, at Refincry B. Becaw of the long latcncy period for most solid armors, a worker was arbitrarily considend 'exposed" to a work category if he was known to have worked at ltast 1 day in the category beginning Is or more years prior to his death. Although the latency period for leukemia may be C0"sidaably shorter. changing the definition of an "exposed" worker by eliminating ttae latency period did not alter the results. The total number of days worked in a specific category was calculated for each yexposcd" case and control. About 40% of the study subjects met thc 'exposure" requirancnts for only one category; some persons had 'exposure" to as many as six categories. The average number ofcategories was two. Statistical Analysis Maximum-likelihood estimates of the rclarive risk . . .dense intuvals for brain tumOr. 5%nachcancer,and- 1 were dculated by -------- I-S procaurr-C =ror ma_mI eu ~ a ~ c ~aa~ta mwitrh ~a l variable matching ratio IMiettmen.;],lOVdr5ds ratios w u e calculated on~yfor andrefincry work categories subcategories ofcategoy 7 in wtiici~five or moncaseS w e n identified as 'exposed." &cause of the skewed disaibutions for total duration worked in any of the various work categories, median durations were reported for thosewho w e n "exposed." Odds ratios wen consided to be stacistically significant ifthe 90% c o a f i i intavai did not indude 1. The satistical significance tcsting wdisnottobeinterprtredinthesaidsense,butistobeusedasaguidetothe dative importance of the numerouscOmpariSON made. Trends by length of employment were exarm*ned utilizing the Mantd extcnsion procedurr ('Mantel, 19631 for those categories in wfiid! the & ratio was 1.5 or grcalcr, the mcdian duration of employment was greater for the case5 than the controls, and the median duration of employment for the cases was greater than 5 8 ThoarPrdd 9 years. Matching was dropped. and durationof employment levels were defined as he following: 1) mit 'ucposed;" 2) 'exposed" but worked less than 5 years in the category; 3) "wrposed" and WOW5 years or more in the category. Staadardited -risk ratios werc calculated for exposure levels 2 and 3. staadardued tothelowest ex- level. RESULTS Table I shows the numberJof casesand controls asata' ud fiomtadrrctincry and PMRS computed in theoriginal mortality study. PMRS for braia Qunor, stomacfi cancer, and leukemia w e n significaatly devated among whites, and a significantly high mortality ratio for stomach cancer was as0 seea among nonwhites Fomaset al, 19821. The m a x i m u m - l i i M estimates of tfie odds ratios for btain Nmorby.refinery work categories arc shownin Table II. Data for Maiatcnanccand Laborsubcategories are shown only for thase in which at least5 casts workcd 1 day. The odds ratio of 1.6 seen among woT1<crs "wrposed" to the Lube Oil category was not scatisticaly significant. The medianduations worked in this category wen 1 year foi the cases and 8 years for tkcontrois. The odds ratio forbraintumor amongworlrersemploycd in the Maintenance and Labor catcgory was only slightly elevated. Within this category, there was an odds ratio of 1.5 for the of pipefitting. but the maiiaa d u r a t h worired by the carts was very brief a d was less than that worlred by the controls. Although a relative risk measurc was notcalculatcd !brcqatry, four cases and only two contmis worked thcn. A ssdstically significant odds d o of 2.8 for brain Nmor occumd among persons employed in the Receipt and M o v e m a t category; however, the median duration of anployma# for the conads was more than four times longathan that for the casts. Then w e n i n d f i-aIualLnbcIs of cases employed in the Chanicd Divisions at ref&= A and B, Compmsim Plant at reflncxy A, or the M o r Unit at refinery B tojustify sqaratcaaalyses. Resuits for stomach cancer by work category are PreJaUed in Table III. An elevated odds ratio for stomach canctr was seal amoag "urposad" Worlccrs in the LUIX Oil catqpy. ~themaiianduratioaofaqioymcatforthccaseswas-. twh that fbr the contmb. An odds ratio of 4.5 amoag pasoas emplaycd in the MaintenanceandLaborcatcgory was statistically sigdkant. Casesanpiayeedin this category had a mcdian duration of 6 yean of employment ampared with 4.5 yean for the controls. 'The magnitude of this msdt couid notbeurpiaincdbyan wusually high riskassociatedwichanyparticularsubcateggcy. Odds ratios f o r d cancer were devatcd for scveealb h b a m c c andLabor~ries,iadudingpipefitting, boilumalcing, painthg. yard labor, and misCeuananls maintenonoe activiw,how- ever, median d u d o n of anplaynaart for the oscs was less than 2 years fix all subcategories ULCcp painting (18 years). Althougtr odds ratios wen not calcuiatcd for the Laboratory category becauseof the smaU number of exposedcascs, thaewas alargcdiffenncebawcen drecasesandammb i a t h e m e d i a n y e a n ~ ( l 2 years forthe cases vs 1 year forthe controls). The odds atio forsanaachcanccrwas also elevated among workers in &e Motor TransportatiOn category. bur the median duration carploycdwas shorn for the casts than for the controIs. .- .cere & f i n d as the :an 5 yean in the *oIy. standardized : I Z d to the lowest I f t +.i i I rom each refinery .in tumor, stomach and a significantly whites [Thomas et I tumor by rcfmuy ibor subcategories The odds ratio of .Is not statistically year for the cases workers employed \ u t d . Within this 4pefitting. but the . than that worked tied for carpentry, d i c a n t odds ratio -* --tip and Move- [ ccontro~swas dficient numbers LI B, Compression .irate anaiyses. . 1 in Table m.An cl" workers in the -2 cases was almost z employed in the s employed in this .red with 4.5 yuvs Ld by an unusually tor stomach cancer &ding pipcfitting, activities; how. I M 2 years for all .*erenot Caiculated -aiCascs, therewas years worked (U .lomadl cancer was .9ry, but the mcdian 'g pcrsom who had .deither rcsult was I8 E -1 c' 0 E m Y4 1 ! ! L tI PI .. statistiqaily significant: however, in both situations the median duration of cmploy- 'mcntfor the cascswasDgrcatcrthan I5years and was longer than that for the controls. In the Mantel extension analysis, ,mdad.md risk ratios (SRRs)for stomach CilllCcT lllcnascd with length of employment among workers in the Lube Oil and Maintenance and Labor categories (Table V). Although the SRR hacased with duration employed in the MaimcnaeCe and Labor category, durationof cmpioymcnt was XKHs u f f i a d y long in any OF subcategory to ucpiain the apparrnt BSSOCiation with employrQultin the Maintenawx and Labor category. SRRs for la,lrania increased with duration of cmpioyment in the Trratingcategory and in bokmakhg. None of the trendsshown in Table V w e n statistically signibnt. Table VI shows results by refincry for two of the Categories showing a trend in Table V. Odds ratios for stomach cancer w e n elevated among Lube Oil caugory %xposed" pasons anployed at r e M a AandB, andtheodds ratiowas SEatistiCauy signifiornt only for d i n c r y A. No odds ratio was EalnllarrA for rrfinery C because of tfie small numben. Odds ratios for stomach cancer among persons "utposcd" to the- and Labor category w e n elevated for each of the dvee refineries but none werc sta&ticaUy significant. ResuIts for leukemia were not analyzed by refinery becauseof the small number of casesemployed at each m.- compared with the pooled controls, M unusual distribraiocls by work category were noted for the cases of multiple mydoma, Hodgkia's disease, mn-Hodgkin's lymphoma, or cancers of the skin. prostate, and paacrrar. DISCUSSION ---- In this study, we examined the relationship between cancer mortality and refinery work categories using ddailcd work histories from three oil ntiIlclies in furatioa of cmployhat for the Controls. (SRRs)for stomacb I the Lubc Oil and ;RRincnased with tion of employment qpwcnt associatioa b for ltukemia innd in boiilermaliag. ishowingatrendin Lube oil category atio was statistically refinery c hrralLcc !fSON 'exposed" to the throe nfureries :re not anal- by Zfurery. IS by work category ~SC, non-Hodgkin'S Y i i 5; ii s i I :mortality and rce oiI ref&= in wry workcategoria 10 19 1.8 25 56 3. I 4 6 20 3 S I .8 TABLE VI. Number Empbycd at Lesst I Day and Odds Rat& fws(ool.chC.aar by R e f h r y for Sdeaed work cntcgOrie3,ls Y a n Latency work cucgory No. anpbycdat l a s t I d8y eyes coatrds odds nti0 90% Confiderrcc imcml 02 Lukoii Refinery A 7 IO 3.1 1.1- 8.8 - -Ref- B Rcfinayc -. 10 2 23 7 1.5 0.7- 3 5 mhlrinrmr#r andLbor RrfiaayA 18 48 3-6 0.6-19.6 21 56 2.4 0.7- 8.8 RcfinayC 8 IS 3.3 0.9- I 1A Texas. Cerorin Iimitations in the nsounxs aad matrodology must be Consideredwhen imtrpraiag resuits. T i a few differtnces in job and deparunental organization b a w m the three refineries oould not be rcsolved in classifying record e m i t s into work catcgorits. Qlanga in rinery op~ratiorrS,~c&oIo~Y, d-mtal omiza- tion, aadjob classifications have varied over time and may have introdud additional dassificatioa inaccuracies. Second, it is rot clear whether our work categoria were truly iadicative of exposure since the numbaofpomtial exposures in any particular category was extensive, and because wwkcfs may have had contact with a specific s u m in more than one cattgory. Third, thue was a greatdealofjob mobility in &ref- and many ofthe study subjcqs were classified as 'exposed" in multiple work ouegories. For example, tbe majority of brain tumor aad stomachcancer ~ t s e s 'exposed" to odrer otegorics witb devatcd odds ratios had also worked in the Mainttnancc and Labor category. On the 0th- hand there was virmally no overlap bawecn 'exposure" to Treating and Boilermaking among the leukemia cases and their matched controls (1 case, 0 controls). Fourth, the number of cases with each form of cancer was nlativeiy small. consequemiy, we knew that the Lilrdihoodof detectinganelevatcdrisk for aparticularcancerwas poor unlcssthe actual risk was very high. Fifth, our comparisons were madc bawan casts and matchedcontrols all of whom worked at oil refineries, so that we may have overmatched with respect to gcncralocarpational errposurcs withinthercfinaymironment FMy,because we arr maicing multiple comparisorrs, a llumbtr of scatistically significant d t s would be expeaed due to chance. Using the work categories defined for this study, elevated odds ratios for brain tumor were sea among OCAW members anployed h the Lube Oil, Rcccip and Mov- and m f i h g ad-; howmr, rhe shona of r n p b v . . .fortbe casts compared to tbc CoIltroLS is Ilot con!&wt with an etiologic role for occupatioaalurposurtsintheseparticularcategomaElons Iftong-termrefinaywork- malen in gaxxal have an elevated risk of brain tumora due to so= diffuse UqroSuFe withia the plam, for urample, polycyclic a x w m t ~hy~drocarbon)sP'( et al, 1981; Schottenfeld et al, 1981). the w i d h p h t cascumtml method would not be effective in unoovering such a risk. Brain tumors have been iaduad cxperimenrallY by cutaia PAHS [Hosobuchi and Ishii. 1967, Zimmamanand Amold, 1941,1943], aad snail excess risks have beensuggestedamnagOttrerocatpationaigroups exposed .- 14 Th0m;rreZli 9" to PAHs. inciuding aluminum reduction worken [MiUlam, 1976, 19791 and motor vehicle examiners [Sternet al, 19811. The most intriguing associations found in thisstudy werc thosebetweenstomach cancer and employment in the Lube Oil and Maintenance and Labor categories. A positive rciationship by length of employment was seen for bothcategories. Excessive mortality from stomach cancer among rcfincry worlccrs has been reported in some d e s plomas d a, 198L; Hanis et aI, 19791 but not orhers -[ et d, 1981; TheriaUit and Goula, 1979; Hanis a al, 1982, Aldcrson and Rrrshton, 19821. None of these studies reported paarms for Lube Oil workers scpaatdy. The bask process in the Lube Oil are8is thc separation of lubricating oil fnwnpetroleumwaxes. Prior to the 1950s. this process involved the refrigeration of lube oil stodcs and removal of solid wax by filter pressing, a pnnxss previously rdatcdto nonmeia- noma skin cancer [Hendricb et al. 19591. About 1950, a solvent dewaxing pmccss replad &e prcviow system. Among "exposed" study subjeasemployedin the Lube Oil category for more than 5 years. eight of ten cases (80%)and only 12of 19 (63%) controls hadWOWin areas producingor packing padfinwax. Most of thestomach cancer caseswho were employed more t&an5 years in the LubeOii area had worked prior to 1950 in arws where paraffii wax was bciig scparatcd from lubricating oils or prrpared for packing. ALthough no expected value was caicuipted,duet stomach cancers were observed among a cohort of 82 wax pressman studid between 1937 and 1956 (Hendricks et al, 19591. Canccr of the stomach has also been associated previously with pnsumcdexposurrs to lubricating oil mists in m d h h g opccatiolrp [Decoufle, 1978; Jarholm d al, 19811 and other occupahns [Okubo aad T d y a . 19741, particulates such as carbon black among rubkt wdccrs [McMichad et al, 1976; M o w n and Nakam, 19761, and with exposures to PAHs among mfm [Hammondet al. 1976). No such relationshiphasbeen seen among other Occupational groups expa& to PAHs including coke oven workers [ReQloni et ai, 19761 and coal gas worlws p o l l et a, 19721; however. a variety of PAHS given d y to rodents induced stomach tumom [Santdorraul et al, 19811. Ttre relationshipbetweenstomachcancer and MaimaranCe d Labor activities may reflect -xe to pcvoleum stock and produds, or to asbestos,Maimcnanee work is generally chammr'& b y h i g h ~ d ~ u r p o n r r c s t h a t o c c u r d u r i n g shutdowns and proMan siatatiacrs. Elevated odds latios forsaomrdr cancer among pipefitters and boil- suggest the ruleof askstosexpsul& Qmiausiy IuJoci- ated with s t o d oiocc~risk [Miller, 19781. Nosigniticaatassociatioruforleukemiaoccumd wicb any spccifiijob category. 'IhehighcstoddsratioaxxlmdintheTrratingcategoy,~alsoaapositive 1-p with duration of anpioymtm. During tradng, pccroleum poduar are -further rdined by rrQIcing the lev4 of aromahc andsulfurconstia#nr d a r e .diwith additives to improve theitquatity. A I t h o u g h ~ e x p o a u l ep, mi- - . .-lY with teukania risk [vigliani and Saita 19641. was an a priori . .hypo&&, rhat4z%poswis naoucgory-specific. The BSJOCtaOODw i t h b o i i d g wasnotstatlstlcaity s i g n i k a a bur this finding is consistent with resultsof astudy of ~ shipyad bodennak#s (Bcwmoa 19801. In summary.ouf study hiled to detect any strong mochtioas becwetn brain tumor risk and employment within refinery work catcgorics. This might result from liitations of the occupational categorizations. smaU numkrs. ot a true abscnce of devatai risk. Because ehe ref- work categories defined here are noc exposure- REFERE .. , 6, 19791 and motor ~riebecweenstomach laborcategoria. A ;JtCgOneS. Excessive .nreponed in some IS&ottenfeld et al, id Rushton, 19821. .parately. The basic 'rn petrolcum waxes. lube oil stocks and rrlatcd to nonmelaIt dcwaxing proctss rnployed in the Lube emly 12 of 19 (63%) Most of the Stomacfi ( )il area had worked mm lubricating oils !.rted. three stomach iidied W e e n 1937 elso been associated whining operations .hub0 and Tsuchiya, [McMichaeI et al. c . among roofers .moccupational xi a al. 1976) and .,Hs given orally to and Labor activities .lestos. Maintenance that occur during :I& cancer among .previously assod- twzcificjob category. a showed a positive ldeum products an -0rwituents and arc .ne exposun. prcvii1. was an a priori It with boiilcrmalring d t s ofa study of .iom buwm brain IS might d t from w a true absence of 2 arc Imt exposuro I: i iI 1. i f 4 I ? Cancer Momdity by Work 43tqp-y 15 specific, workcrs urposcd to particular chemicals d d have elevated risks of specific cancers regardless of the work categories to which they were allocated. A siight asmiation was seen for leukemia risk among workers in the Treating category, and stomach canccr risk was elevated among maintenance men and workers exposed to lubricating oils and paraffin wax processing. Fu~thcrrcscLcch involving iarger study populations and more detailed e x p u n ciassifications is neededto clarify the role of occupationaldetcrminantSof these tumors in the peaplcwn industry. .:a I Organ< Poison ClCnica Y. h a r JMROOUC