Document 2Jz9qq2DebwezVBjwmjMqwKG7

7 British Journal of Industrial Medicine 1986;43:6--17 An epidemiological study of petroleum refinery employees 0 WONG.' R W M O R G A N , ' WJ BAILEYZ. R E S W E N C I C K I , ' GATHLEEN CLAXTON,' AND LEEKA KHEIFETS' I An 4 thc < codir Th lowit sex. 1 date of dc somc aftcr ABSTRACT A cohort study of 14 I79 current and former Chevron USA employees at the Richmond ____and El Segundo, California, refineries was conducted. The cohort consisted of everyone working at -___I- either refinery for a minimum of one year. T h e observed mortality of the cofiort, by cause, was com*red-Gth the expected biGG&%itheUnited States mortality rates, standardised for age, race, sex,_andglendZGie7ATaI-ysSbj r c f i n ~ P o 6 - ~ e g o ~ ~ ~ ~ ~ d a t e *em- pdlo~ymt einot,oanfd latency were performed. cor the entire cohort, mortality from all causes was 72.4% of-that expected, a deficit that was s t a t i s t i c a l l y ~ n i f i ~ t X- di_-d- -*d i t ~ n ~ ~ f i ~ e m o r t a l i tw ya s f o u i d G F a l t T o o l cancer comb?ined, di-- gestive cancer, lung canccr,~ea~d~m-se-.non-malignant respiratory disease. showed a pattern of diseases increased orthe d i p ; e s risk suggestiveof t a i -v e- t e m possible re l,_a_dtcioan_~to_~ a~nno ~y ccu ~ pa y ti m on p al h o p ocianecteri exposure. T h e c excess appears confined to cancer of lymphatic tissue (not Ieukaemias) at Richmond, and only among those hired before 1948. A follow up case analysis of the deaths from lymphatic cancer failed to identify a common exposure pattern. Several epidemiological studies of petroleum refinery workers have bcen conducted'-'0 (Divine BJ era/, unpublished report). Although these studies differed in size and design and covered different periods they showed revcral canccr sites with significant cxcefses among individuals exposed to petroleum and its byproducts when compared with the general population or with non-exposcd employees. The canctr s i t a that have bemimplicated in more than one study include cancers of the respiratory system, stomach. lymphopoietic tissues. and, perhaps. brain. Although some of these studies have certain limitations, either because of their design or because of the particular data source available to the investigators, they have suggated possible health risks among refinery employees. To investigate whether these risks existed among its employec~C, hevron undertook a long term mortality study at its Richmond and El Scgundo, California, refineries. Materials and methods DATA COLLECTION The cohort consisted of all individuals who had completed one year of work by 31 Daxmber 1980 and who had worked for at least one day between I.January 1950 and 31 December 1980 at the Chevron refineries in Richmond and El Segundo. California. The cohort was constructed by using employment records at the two refineries, corporate offices. and other Chevron record storage facilities. For efficiency,only the first and last jobs (job titles. locations. and dates) wen coded. By not coding the entire work history, it is possible to miss specificjobs of high risk. Ncverthelcss, we thought that this would correctly identify most subcohort members yet offer considerablesavings in both time for completion and cost ofcoding. I! was also rccognixd that should thc study generate enough interest in certain causes of death. any further investigation would require either a detailed casc analysis or, ultimately, a casecontrol study (with complete job coding of OLKS and controls) if feasible and warranted by theavailable data. A third, less likely, possibility would be a reanalysis of COH We I qual pletc wa5 SpCC sour# pen< empl persa mad ecs acti\ from FI tap SCICL (196 uals (3.8' Rich thew but 1 they rnicr micr miss# OCCll we d ascc: VIT It w. who milit 19%I betU uscd inclt~ (1' low (21 tern f (3 An epidemiolbgkal study Of PetrO~eumrefinery employees the entire cohort, but after adding complete job D E A T H C E R T I F I C A T E A C Q U I S I T I O N coding. We retrieved death a r t i f i c a t a for 2242. or 9742% of The employment records generally included the fol- those identified as dead, most of whom had died in lowing information: name, social security number. California. All death certificates were d e d by a sex, race, date of birth, date of hire (or service date), trained nosologist according to the 8th revision of the date of separation, and an employment history. Dates International Classification of Diseases by underlying of death and death certificates were available for cause of death." some individuals who had died while employed or after retiring. STATISTICAL PROCEDURE The most common summary index for assessing the risk of death in a cohort study is the standardised COHORT VERIFICATION mortality ratio (SMR).The basic unit of computation We investigated the possibilityof using social security is the number of years each employee was followed up quarterly reports (Form 941) to verify the com- (person-years) from the anniversary date (one year of md pleteness of the cohort but found that that method x r v i a ) or 1 January 1950(whicheverwas later) to the gat was i was not practical because these reports were not end of the study pcriod or the date of death (whichspecific to locations within the state. The best data ever was earlier). For those "lost to follow up," source for verification within the company, but inde- person-years of observation were counted up to the a=, and that i ! I pendent of the source of data collection, was last date of contact, which was usually the termi- employee year end tapes from the antralised nation date or the last driver's license renewal date. personnel/payroll system T h e computer tapes are Thereafter, these individuals were taken out of the WaS ! made at the end of each calendar year of all employ- analysis. Thisamounts to assuming that the mortality i ecs who had had some employment or retirement experience of thcx individuals was similar to that of I activity in that year. Year end tapes arc available the rest of the cohort. Each year (or fraction thereof) T h C from 1962. contributed by an individual worker was classified by mly I For verification purposes, three sets of year end age, race, sex. and calendar year. and these penon- iiled , tapes (1965-6, 1970-1, 1977-8) were randomly years of all workers were then summed up by age, selected within each of the t h m six year periods race, xx, and calendar year. The United States (1962-7, 1968-73). 1974-9). A cohort of 7097 individ- national agecaw-race-sex-specific mortality rates uals were generated from thcsc t a p . Of these, 279 for five year periods from 1950 to 1980 were applied (3.8%) were not included in the cohort (183 from tothose person-years to obtain the number of deaths Richmond and 96 from El Scgundo). About half from a particula'r cause to be expeacd from an qual thesc ( <2%) were found to be eligible for the cohort nurriber of person-years of the same race and sex and but were missed in our data collection, either because similar in age and calendar year. Icom;O and I Janievron rornia. pent s, and 1 1 i i they were not present in the files at the time of microfilming or were inadvertently omitted from the microfilming. The few potential cohort members m i d was not over represented in .any particular occupational department or file storage facility and we do not believe. therefore. that any bias in cohort ascertainment would bc likely to result. SMRs were computed by expressing the actually observed deaths as percentages of the expected. An SMR higher or lower than 100 indicated an exccss or a deficit in mortality. The deviation from 100 was tested to determine whether it was statistically significant at the 5% and 1% Ieveh. The actual com- putation was performed using a standard computer pro gram. 1 titles, ng the ic jobs would :t offer >n and uld the lses of :ather :ontrol d con- [a. .A of I i I i I VITAL STATUS FOLLOW UP It was ncrrssary to determine whether those workers who had terminated their employment (including terminces, transferred. and retires) before 31 December 1980 were still living or whether they had died bctwecn leaving and the end or the study. Sources used in the vital slatus follow up of former employees included: ( I ) Social Security Administration Vital Status Follow up. (2) California Automated Mortality Linkage System (University of California, San Francisco), and (3) the California Department of Motor Vehicles. RtsultS DESCRIPTIVE STATISTICS Altogether 14 179 individuals who satisfied the cohort Table I Distribution by sex und ruce o/all cohort members Richntond and EI S e w d o refineries Whiic Non-white Unknown Total IO 127 7111 2 656 115001 508 51 I I6 678 10635 771 2 772 I4 l i 9 !-- 8 tVorrg. Morgan, Buifey. Swcricicki. Cfuxfon. und h.)ie~l~i/~.ts T;ihlc Z Di.vtrrhutiori h.v ycur of birr 4ull cohnrt wirmht*r.vut Richmond cmd U Sgundn rr/;nrrirs Yrar of hire Richmond Frcqucncy Y. El Srfundo Frrqwnry % Both r e j w u s Frrqurncy % < IPZO IY?W 4 1925-9 1930-4 .t:z1935-9 1950-4 1955-9 I9604 l96S9 1970-4 1971-9 Total 2SR 229 I37 VI 356 7% 1788 9R4 946 762 847 861 972 898I 2.RR 2.55 I-53 1.01 3.96 8.35 19.91 IO-% 10.53 8.48 9-43 9-59 10.82 100-00 CS 133 II 5 77 167 43v 837 631 S23 375 517 67I 648 5198 1..2_s_ 2-56. 2.2I I-48 3 2I 845 16-10 12.14 10-06 7.2 I 9-95 12.91 1247 1OOa 123 362 252 168 523 I189 2625 1615 1469 1137 1364 IS32 1620 14179 2.211 349 8.39 18-51 11-39 10.36 802 9-62 10.80 11.43 100.W criteria were included in the analysis. Table 1 gives mation from death certificates was not used and for their sex and race distribution. Table 2 shows the dis- the analysis, those individuals with unknown race tribution by year of hire for the two refineria and for (2772) were assumed to be white. the entire cohort. Major hiring at the two refineries In the study 2292 deaths were identified. Death occumd in the late 1940s. Almost 40% of the cohort artificates were obtained for 2242. The SO deaths were hired before 1950. thus providing a possible without death certificates. but with dates of death, latency of up to 30 years for this group. were coded as cause unknown. These 50 deaths were Table 3 shows the distribution by duration of included in the calculationofoverall SMRsbut not in employment for the entin cohort. Over 47% of the any cause specific SMRs. cohort was employed for 10 years or more and 33-5% were employed for less than five years. The average ANALYSIS OF T O T A L COHORT A N D BY REFINERY duration of employment for the cntirt cohort was Table 5 shows the observed deaths and SMRs by about 14years. cause for each refinery and the total cohort. For both Table 4 shows the vital status of the cohort on' refineries, the total mortality was significantly low 31 December 1980. At the end of the study period, (SMR = 74-1 at Richmond, 69-2at El Segundo; 72-4 I 1 177 (7843%) were still alive. A total of 2292 indi- for both) by comparison with the United States viduals had died during the study period and death national mortality experience. For most causes of certificates were obtained for all except 50 {2-l8% of death, the SMRs at the two refineries were similar. all deaths). At the end ofthe study, the vital status of and our discussion will focus on the total cohort. 710 individuals (501%) m i n e d unknown. Mortality from c a n a r of all sites combined for the total cohort was also significantly lower than the CAUSE SPECIFIC MORTALITY ANALYSIS The race of 2772 individuals was not available from their employment records. T h e n were, however, few Table 4 Diwributwnby vital status (on J! Decem&r 1980) of all cohort members known non-whites (772) among the Chevron employ- ees at the two refineria included in this analysis. In order not to bias any racespecificanalysis, mce infor- Table 3 Distributionby duration of emplojrnenr of all cohort members at Richmond and El Segundo refineries &mplo)7nrncduration (yrars) Ficpvnwl % (5 5-9 10-14 15-19 20-24 25-29 30-34 3s39 340 Total 4753 2752 I417 823 837 I I94 I292 724 387 14179 33-52 1941 9-99 5.80 5-90 8.42 9.1 1 5.1 I 1-73 100-00 An c p -Tahlc Cauw , - All CIII Inkti\ All a n Ccamnc c`. C. C. c. C: C. Can< C. C; Cam Cam Cam caflc Cam Can< Gn. ". H. LL 0 Benign Diabet Discas, Diuast Discas. arc Ane Vas Non-nPnc1 Emr DiscasCirr: Disc& Senilit. Accidc Am -Mol Suic .Sign!, tSip Unitc from phar? (SMF P<P for sc and t Sli* were lYmP, 126.6 (SMI thc R 1344 rctict An epidemiological study of pe~lroleumrefinery employees Table 5 Observed &thc and SMRs by causefor total cohort 01Richmond and EI &gunah re+iu l-u'ural Obi SMR RUU - Obs S M R All a u s c s lnf-tw & parasitic d i x a s a (000-1 39) AU un- (140-209) Gncnof buaul cavity & pharynx (14&149) Gnar ofdigcrtivcsysum (ISO-159) Gnccr of oesophagus ( I SO) Canacr of stomach(151) Canaxoflarge intcrunc(153) Cancer of rectum (154) Canaxof Jim( I 5s156) Gnarof pa- (157) Gnarofrespintory rystcm(160-163) C a e r or hrynx (161) Canar oClung (162-163) Canar of skin (172-1 73) Cancer of ( 174) Cancer of prostate (185) Cancer of tatir(186-187) Canccr of bladder (188) Canar of kidney (189) Canar of bnin and CNS (191-192) Lymphatic k kmatopoictic anccr (200-209) Lymphovmtna k rcticul-oma (200) Hodgkin's disease(201) Leukaemia dr alcukaani. (204-207) Other lymphatic tissue a- (32203.208) Benign neoplasms (210-239) D i s k t a mellitus (2W) DiwzraofMOod(28&2119) ~ i ~or ta-0- ~ srp ~ a n(320-389) Disuses darcuhtory ~ y ~ t c(3m90458) Chronic rheumatic hcandi- (393-398) Anerioukrocichean~(4IW13) VrvuLr ksionr ofCNS (43M38) Non-malignantrespintory dr- (-519) Pneumonia (480-486) Emphysema (492) Disuses of di alive system (520-517) Cirrhosisofliver (551) Di-nf genitourinarysystem (58M29) Senilityk illdefined condi!ionr (780-799) A r d m t r poisonin- & v l o k (EEOO-E998) Acxirlcnts (XUO-y.19) Motor vehicle accidents(810-827) S u i d e (Wtr-PSU) 2292 I4 462 9 125 I2 26 40 18 6 20 135 9 124 IO 3 34 2 12 13 22 64 I7 4 2m2 5 17 6 12 I228 I9 924 I58 105 35 26 97 60 22 IO 213 I46 75 nu 72.4t 37.7t 15-7t 47.0- 7l-6t 83.5 762 71.4. 89-1 49.0 5929 68-7t 99-6 66-9t 92.5 101.8 78.4 53.0 62.4 87.7 125-9 105-0 126-6 52. I 88.3 141.6 61-5 37.lt 85-2 47.6t 73.6t 61.1. 79.3t at 55.7t 494t 53.4t 64.7t 79.0 494t 27.2t 79.2t 71.3 77Q* 127.6' IS07 5 302 7 85 8 21 27 I3 2 13 82 6 74 7 2 23 2 10 9 IS 46 13 3 I4 IS 2 I2 4 10 801 8 605 Ios 71 21 17 68 41 IS 7 I62 9u SI 56 74-I t an.7stt 57.2 76.ot 86-4 95.3 7s-7 100.5 25.4- 6&3 65.6t 104.0 62W 101.2 I27.2 82.6 81.1 81-6 95.2 134.5 118.0 151.3 60.4 87-8 167.0 38.4 404t 88-2 61.8 14.9t 40.1t 8I.2t 684t 594t #It 55.1 t 70.7t 84.4 52-3t 29.3t m9t n.9t 80-8 126.2 *Signifrmt at 0.05. tSignifiant at 0.01. 9 El scgvub Obs S U R 78s 69-2t 9 69.3 160 n . 3 t 2 29.0 40 63.7t 4 78.1 S 41.4. 13 64.0 5 68.8 4 91-7 7 57.3 53 74. I 3 91.9 50 3 74.0- n.I I 72.8 I1 0 -70.9 2 28-7 4 74.5 7 110.7 18 81.9 4 824 I 364 8 89.1 5 97.3 E;+3 102.5 5 2 2 22.2. 427 71.4t II 98.4 319 759t 53 63Qt 34 49-8t 14 55-5. 9 50-5. 29 Y Q t I9 69.5 7 45.39 3 23.2t 81 761' 47 & I t 23 6Y0 32 1300-3 United States experience. n e deficit came primarily canccrs; none of thesc SMRs was statistically from the following sites: c a n a r of buccal cavity and significant. pharynx (SMR = 47-0. p < 0.05). digestive system Mortality from c i n a r ofthe larynx (SMR = 99-6). (SMR= 71.6, p < 0.01). and lung (SMR = 66-9, skin (SMR = 92.5). breast (SMR = 101-8). and kid- p C 0.01). Non-significant deficits werc also observed ney (SMR = 87.7) for the total cohort was similar to forsevcral othcr sits. including mnar ofthe prostate that expected. and bladder, Hodgkin's discax. and leukaemia. Both benign neoplasms and diseases of the blood Slight. but non-significant. incrcascs in mortality showed a non-significant deficit (SMR = 61-5 and wcrc found for c a n e r of the brdin (SMR = 125.9). 85.2. rcspcctivcly). Statistically significiint mortality lymphosarcoma and redculosarcoma (SMR = dcficits were observed. however, for many non- 126.6). and othcr lymphatic tissue canccrs m;tlignant causes: disuscs of thc nervous systcm (SMR = 141.6). The ~ X C C S S ~ Scamc primarily from (SMR = 47.6, p < 0.01) and circulatory systcm thc I<ichmontlrcfincry. wlrich cxpcrionccd an SMR of (SMR = 73-6. p < 0.01). nonmalignant respiratory 134.4 for brain cancer. 151.3 for lyinp1ios;lrcottia atid d i r a ~ s( S M K a 55.7. p .( 0 4 I), d i m a c s d tllc n.ticulusr\rcoma, i d ,167.0 for othcr lymphatic tissuc digestive system (SMR = 64.7.p 0.01), and gcnito- 10 Wong. Morgan. Bailey. Swencicki. Claxfon. and Kheifeis Ta blc 6 Ohstrved clcarhch.v cause and S M R sfor all coharr n t e m k s in laborarory. mainrenonce. and operarinK lmarionc at Richnrond crrrcl El Sc*g~trdcri.ji,rc*rirs Cousr of akarh (8rh ICDA J All a u y I Infectivek parasiticdiseases(-1 39) canam ( I4G209) 'lk-rof bucalavity k pharynx (140-149) Canarofdigcrtivcsystcrn ( I S 1 5 9 ) Cancer of ocrophagur ( ISO) Cancerofstomach( 151) Cancer of large intestine(153) Gnarof rectum (154) Canar of lim(I S S l 5 6 ) Cancer of pancreas(157) Gnon of respintoysystem (160-163) C a m of luynx (161) GnafO f l u g(162-163) C a mof skin (I72-17)) C a m of b(174) Gnarofprosutc(185) Gmoflutir(186-l87) -(m)Cancerof bladder(188) Gnar of kidney (189) GnocrofbninmdCNS(191-192) Lym*cickhtopo~Mccr(200-209) Lymphosammu k Nculov Hodgkinfdisasc(Z0I) LulucmhkakuL.rmu(W207) otha r~mphrlictis= tlllocn(mol203.208) Benign neoplasms(Zl0-239) Diabcca mellitus (250) Disases of blood (280-289) Dwur of nervous system (320-389) Di- ofarculatory system (390-458) Chronic e u i c hacd- (39S398) Artenoulerot~heandisease ( 4 1 0 4 3 ) Vascular lesionsOCCNS(43W38) N o n - n u l i p n l Rspintory disuse (460-519) Pncumonu ( W ) Divpsaofritourinary system(580-629) Senility k i ddindcoaditionr (780-799) Acddmts. poisoning. k violence (uuw)-E998) AcEidenu(800-949) Motorvehidelocideau(810-827) suicide (950-959) *Signifiant at 045. tsigniht at 0-01. Laboratory Obs SMR -13* 74.lt 0 -29 81.7 0 7 71.2 -0 -0 2 63.2 -4 356-4 0 I 51.2 -3 U.lt 0 3 246t -2 M0.I 0 -3 133.1 0 I 95.4 2 W.8 I 90-1 6 l65Q -2 218.6 0 I 68-8 -3 357-4 0 -2 77.3 0 I 66.0 -71 76-8. 0 57 86.5 6 18-8 5 47-6 I 263 2 72-4 4 44-5 3 62.4 1 42.1 I 46.6 I 7 81.7 7 53.4 1 15-90 9 205-7 Maintenance Obs SMR I329 I2 266 4 70 7 IS 17 13 5 I2 80 7 72 7 I m 2 7 7 14 36 IO 2 13 II 2 7 3 6 719 9 541 93 61 21 12 58 34 II 6 138 90 41 44 n.2t 566.0. 75Qt 35.8. 684t B2.6 75.2 524t I10.6 70.7 608 @It 131.6 65-9t 112-5 SI50 769 93. I 61.7 80-9 139-9 102.3 129-2 45.9 90.2 133-7 43.1 26.St 74.3 41.6. 73.6t 51.0. 794t 67-Ot SSQt N.6t 41.6t 67Qt 77.7 42.8t 28.lt mst 780. 76.7 113G Operating Obs SMR 817 7PSt 2 18-2t I63 82-5. 6 96.9 -36 6 3 4 t 0 IO 91.8 13 71.2 5 76.6 2 51.6 n4 36-50 83-7 4 I360 50 82Q -2 57.I 0 7 18Q. I 852 6 93-9 3 62.3 7 127.5 28 143.1 7 1643 3 127.4 10 124.2 7 IJI.5 3 118-6 4 27.1t 5 219-5 . 5 62.5 42I 77-zt 2 21-St 329 86Qt 46 59Qt 39 62.5t IS 64.3 9 55.4 38 79.I 26 106.6 5 35.90 2 16.9t 96 98-7 59 92.3 31 104-6 33 IM-5' urinary system (SMR = 49.8. p < 0-01). senility and ill defined conditions (SMR = 27-2, p < 0.01). and accidents (SMR = 71-2, p 031). Among all the causes ofdeath examined, mortality from only one cause. suicide, was significandy raised (SMR= 127-6. p < 0.05). Both refineries showed an equal amount of increased mortality from suicide. ANALYSIS OF WORK LOCATION As mentioned eadier, the locations as well as the job titles for the first and last jobs w e n coded. The locationswere'dassifiad into scveral broad categories. Three of the major categories wuc: Operuhg-Light oil. lube oil, thermal cracking, catalytic cracking, chemical and packaging. Maintenance-Field maintenance, shops, quip rnent i n s p t i o n , labour, and engineering. Luborarory-Analytical laboratories, chemicallab- oratories, physical laboratories, inspection, and tech: Neal. In this analysis cohort members were grouped by location (either the first or last job). Some of the members were in two. groups. This analysis repre- sented an attempt to identify any association between gencral work locationswithin the rcfinerics and mortality. Detailed analysis (such as by duration at a ccrtain location. combination of location. and job title) was deemed inappropriate, s i n a only the first and last An t Tah Coir - All, Infct All L C. C. C.J, t cd C.8 C. G C3 G G L! I I &ni. Diah Disc. Disc., Dix. 0 Ar Va Non Pn En Disc., CII Disc. Send Am, AL M. SUI *Sip tSigii job 7 em \ loci tor! con Jew sigr of I vel1 laic P clay An epraemrologrcal study OJ p e t r o l e m reJinery empioyees ile 7 O b y m d & a h by cousc and SMRr for the total cohort ai Richmondand El Scyndo r./incric.cby latency ~~ 11 All uuxs Infective& paruiticdiscasa(OO0-139) Allcancm (14O--X9) Canccr of bucul a w t y & p b q n x (140-149) Canccrofdigc~ive~yru(mISO-159) Cancer of -plugus (ISO) Cancer of stomach (151) Canocr of luge inustinc (I53) Cancer ofmctum (IU) Cancer ofliver (ISS-I 56) Canom of ( 157) Canccrdnzspimory syslan (160-163) ~~~1orbrynx(l61) Cmardlung(162-163) Cancer oCrkin(I72-l73) Canccr of br+act (1 74) Canccr of prostate (I85) C a m of laxis (186187) Canccr of bladder (188) Cancer of kidney (189) Canccrolbnin a d CNS(191-192) Lymphatic& hacnucopoieticam(200-209) Lymphosarcoma k mid- (ZOO) Hodgkin's disease (201) Leukaemia& .kukaania(204-207) Other lymphatic tiuuc a- (202.203.208) Benign neoplasms (210-239) ' Dubeta mellitus (250) . Di- of blood (280-289) Di- of nervous system (320-389) Divua ofarculntoryryuaa (390-458) Chronic rhcumalic hean disease (393-398) Aftcriaoclerolichandisav(41M13) vuculu ksion,of CNS (430-438) Non-malignant respintorydisease (460-5 19) hcumonia ( W 6 ) (42Dizzr sysccm (20-577) cirrtrosis O K (55I ) h s a of genitourirury systCm (580-629) Senility k illdcfi+ mdi!ions (780-799) Accident& poisonings k noknrr (E800-EW8) Aaidcnu(800-949) Motor vehicle racidmu(Sl0-827) Suicide ( 9 S 9 S 9 ) *SigniSant at 0-05. tSignifiant at 0.01. 169 2 21 0 2 0 0 2 0 0 0 8 0 8 I 0 0 0 0 2 2 2 0 0 I I I 0 0 I 43 2 33 5 2 I 0 8 4 4 I 79 58 32 18 388 7 6 3 I735 n2 23.20 IO 79-7 369 I 32.4 8 14 55-2. 109 I 48.6 11 94-3 S 909 21 794 2 27.5- -4 131.8 0 36 789 I4 874) 6 63.4 2 41.6 I 8 65.5 l5 914t 102 63.4t 2 151.5 7 95-7 23 89.9 93 61-It 2 83.4 I 0 -117.1 -I 904 0 7 .9-7.-9 2 34 I I2 71.11 3 125.2 8 69-0 7 164.3 13 119.0 12 107.3 SO 115.7 5 180.7 12 128-7 1 46-1 3 82-4 I 23.2 20 111.1 --5 263-3 0 0 14 121.9 4 17 I 88.5 5 I IPS IO I 8 6 8oot 999 5 62.3 I 2 I52 92.6 139 12 4S2t 141 I2 54.40 91 2 21-80 32 7 145-3 19 20 63-1- 69 I S 8 3 5 41 81.2 3 41.1 IS 1 14.90 8 71 83.7 93 41 73.8 47 19 71-2 23 30 ,148.7- 40 jobs were coded. Table 6 shows the mortality of the subcohorts of employees in laboratory, maintenance, and operating locations. Among the 844 employees in the laboratory locations. the overall SMR for the two refineries combined was 74-1 which was significant at the 0-01 level. In addition. three causes of death showed a significant dcficit: lung cancer (SMR = 26-6). d i x a x s of the circulatory system (SM.R = 76.8). and motor vehidc accidents (SMR = 15.9). None of thc calculated SMRs was significantly rdixd. About half the cohort members (7838) w e n classificd as having worked in maintenance locations pincanriy lower Inan expected (table 6). Significant :ality deficits were found for all Canars -(SMR = 75-0). digestive cancer (SMR= 68.8). lung cancer (SMR = 65-91. diseases of the circulatory system (SMR 73-6). non-malignant respiratory dis- ease (SMR = 55.0). and accidents (SMR= 78.0). Slight, but non-significant, mortality excesses were detected for cancer of larynx. cancer of brain, lymphosarcoma and rcticulosarcoma. and other lymphatic tissue caners. The 5510 cohort members classified as operating '(-I-llidulfacturing) personnel showed significant deficits I H I ! I I. :1; .c 12 WonK. Morgan. Bailey, Swencicki. Claxton. and Kheijets An e: wcrc IClWC< buration of employmenr from deatl <5 years S-I4 years > IS 3uors (sign Na * All causes lnfcctivek parasitic dirara(OO&139) All u n m (140-209) Cartar of bucalavity k pharynx (140-149) G n a r of digestive system (I5159) Gnnofoerophays(150) G n n o f stomach (151) Gnn of largei n t a t i n ( I 53) Cannofratum~154) ClnanofIim(l5SlS66) Gnnof pa- (157) G~oCrnpintorysyrtan(1~163) cUrrrofluynx(161) Cancer of lung (162-163) Gnn of skin (172- 173) G n a r of bmast (174) Gnn of prostate(185) G n o c r of I d i s (186-187) G n c c r of bladder (188) Gnan.of kidncy(l89) Gnn of bnin and CNS (191-192) Lymphatic k hmucopoiccic cancer (200-209) Lymphosarcoma k reticuiou-a (200) Hdgkin'rdi- (201) Leukaemia k akuluemia (204-207) 0thlymphatic tisw culocn(202203,208) Benign ncoplisms(210-239) Diabetes mellitus(250) Diseasa of blood (280-2119) Diwuerof ncwous system (320-389) Dirara of arculrtwy s y l l a n (390-458) Chronic rtKumUic hartdiscuc(393-398) Ariaionlero(ic haridisease (410413) Vascular luionr d C N S (4J0-438) Non-malipant rnpirrtoryduarc (raCrSl9) Pneumonia (180-486) h p h y v n u (492) D i s a s a d d i IIVC s y l t a n (520-377) cinhodroKvet(33I ) or genilOUnNrys y s m (580429) Senilityk illdefinedcondi+nu.(l8&l99) Accidents. poisonings. k no*ncc(EEOO-E998) Acddcnu (80&949) Motor vehicle &IS (810-827) suicide ( 9 5 9 5 9 ) Ohs 262 2 41 0 7 0 0 4 I 0 2 21 0 21 I 0 0 I 0 4 3 3 0 0. 0 3 1 0 0 I 84 2 64 13 6 I 2 17 14 3 I 19 56 30 20 SMR O h 8@lt 45.1 -84.5 --53.8 93-6 - 144 -121-5 128.4 --58.I -an4 284. I 105.2 --3 8 9 - 219-9 --89-1 24.9 71.lt 48- I 76.2962 44.79 17.7Bo. I 88.4 I180 722 18.5 88.2 94.7 92.3 I 19-9 525 I 89 I 17 3 2 5 4 0 2 34 I 33 4 I 4 0 0 3 8 I1 3 I 2 S I '3 2 2 256 4 2~2. 23 2.5 8 6 27 19 4 5 8sz3 24 21 SMR 18.9t 55.8 72.9t 2.5.4 mot 108.1 28.9 47-5 -101*1 30.8 89.9 57.3 92.8 143.2 93.2 --65. 1 97-2 168.2 18-3 94-3 41.1 36.2 178.0 67.4 32.2134.6 31.8 80-It 44.2 90-6 n.9t 14.3 59-3 79.7 14.1 93.9 40.8 58-9 870 81.3 77.5 I 26.6 Ohs 1505 6 326 8 101 9 24 31 l 63 li 80 m8 5 2 30 1 I2 6 II 50 I4 3 20 I2 3 14 4 9 888 13 658 122 14 26 33 27 IS 4 81 38 20 41 SMR 69-2t 27.2t 15-3t 59-4 79. I 86-1 96-6 15-3 81-6 67-6 6a-8 566t 12I 4 52-4t 79 0 MQ.8 83-8 74-4 79 3 . 518-I It 1.5 12'8.1 1612.7 815.0 I2'24 1210.6 ' 611.2 413-Jt 1 6 7 7'2.6 7'6.6t 6i7.W Ji2.M Ji0.4t 4M 6 t i62t dd0t 4LP7t I7-4t cibst 46 - 4 t fil-4. 1312.6 cxcq thele systc dent ANA Tab1 SM F empl dowi nituc wan canc n lymr For sma; mall retic she\ obsc durr othi Iyml AN: The exp\ XVC kaei darc hist1 dart 50 r Ir hire ordL for all caws (SMR = 79.5). all cancers (SMR = 82.5). digestive cancer (SMR = 63.8). cancer of prostate (SMR = 48.0). diseases of the circulatory system (SMR = 77.2). and non-malignant respiratory disease (SMR = 62.5). Non-significant excesses were found for cancerof larynx, lymphosarcoma and rctic- -ulosarcoma, other lymphatic tissue cancers, and dis- eases of the blood. A significant excess was detected for suicide (SMR 150.5, p < 095). ANALYSIS BY LATENCY Long latent periods a x usually required for chronic d i v to develop so that it may be more appropri- ate to examine mortality experience only after a certain lag period has elapsed. Table 7 shows the observed deaths and SMRs by-cause for the total cohort by latency since hire. In general. persons followed up for lcss than IO y t a f~rom hire date had e ~ c e p t i ~ ~lolwl ySMRs; Among those followed up for 1049 yean, the only significant SMR was for suicidc (SMR = 148-7, p < 0.05). In the samc group SMRs for uncer of thc m u m (131.8). lymphosarcoma and rcriculosarcoma (180.7). othcr lym- phatic tissue cancers (263.3). and emphysema (145.3) 1 I I red L SM befc afte, lym. 149 dea i (Shi bcra OhS six t OCCl non were raised, but none was significant. For those fol- lowed up for m6re than 20 years. no significant excess from any cause was found. In fact, most causes of death were either similar to or less than expected (significant or otherwise). ' No obvious upward trend by latency was found, except for cancer of the digestive system. Neverthelcss, when the subcategories within the digestive system were examined, no consistent trend was evident. ANALYSIS BY DURATION OF EMPLOYMENT Table 8 gives the observed number of deaths and SMRs. by cause, for the total cohort by duration of employment at the refineries. There was a slight downward trend of overall mortality, but the magnitude of the trend was small. More prominent downward gradients were observed for lung cancer, kidney cancer, and cirrhosis of the liver. The only obvious upward trends were those for all lymphatic and hacmatopoietic cancer and for suicide. For suicide, however, the magnitude of the trend was small. When the subcategories of lymphatic and haematopoietic cancer were examined. lymphoma and reticulosarcoma. Hodgkin's disease, and leukaemia showed a slight upward trend. albeit no death was observed for any of thcse t h m causes in the shortest duration of employment ( c5 years) group. On the other hand. there was a downward trend for other lymphatic tissue c a n e n . ANALYSIS BY HIRE DATE The'refinery workers in the study were potentially exposed to benzene. which has been implicated in '*several previous studies for increasing the risk of leu- kaemia and lymphomas." Occupational standards for benzene have undergone several changes historically. In 1947 the recommended benzene standard (8-hour TWA) was lowered from 100ppm to 50ppm, and was further reduced to 35 ppm in 1948. In this analysis the cohort was divided into those hired in or before 1948, and those hired after 1948, in order to assess mortality risks before and after this reduction. The most striking contrast was in the SMRs for lymphatic and hematopoietic cancer before 1948 (120.9 based on 58 observed deaths) and after 1948 (46.2 based on 6 observed deaths). For lymphosarcoma and reticulosarcoma. the SMRs were 149.6 (16 observed deaths) and 36-6 ( I observed death), respectively. All 22 deaths from leukaemia (SMR= 110.2) occurred among those hired in or bcforc 1948, and no deaths from Icukacmii~wcrc obscrvcd among t h o x hircd aftcr 1948. Similarly, all six deaths (SMR = 103.6) from diseascs of the blood occurred among those hired in or before 1948 and nonc among those hired after 1948. 14 I Year of death 1970c 1960- 1950194019301920- Wong. Morgan. Bailey, Swencicki. Claxton. and Khei/ers I1980 -1970 - 1960 - 1950 - 1940 -1930 -1920 / i. I I U v ii I' I r c 1 I 1 I c c I I I 1 Employment histories of 37employees who haddicdof non-Hekin's iympha8iccancer. availability of specific exposure information among these deaths a thorough rcvim of the complete work histories of the 37 deaths from non-Hodgkin's lym- phatic cancer (ICD200, 202, 203, 208) was under- taken. The review was conducted by members of the m@icalindustrial hygiene. and safety_dtpartmq@, who were knowledmable about both past and c u m n t opcraaons and job task at the refineries. The figure shows the job locationsof the 37 deaths by year of him, ycar of death, and ICD codc. Thc figure was developed tofacilitate the identification of common work locationsby specific ICD category and calendar time. By visualising or drawing a line through a single year or band of time (a five year period across the graph for example), it is easier to identify any dusters by work locations. No obvious similarity in job assignments was evident from this figure. At first glance &e numberof maintenancejobs appeared somewhat high in proportion to other work locations; 22 of the 37 (59%) had ever worked in maintenance a t some time during their employment at the plant. Excluding caxs 9 and 33, who worked for less than one year in maintcnancc. the percentage becomes 54% (20/37). At Richmond this percentage b 50% (14/28). These percentages werc not unusual because typically close to half the refinery workforce Table IO Obseweddmhs. e-vpecteddeaths. and SMRsfar malignant cancer offhebrain and CNS (191-192). benign brah hmrours (225.0). and brain tumaurt. M ~ U ~mCp e c ~ e d(Ua)for total cohort at the Richmondand El Segwtdo re$neries Cmur o/&ath (8th ICD) Obvmd&achr Malignant anar OcbninUd- CNS (191-192) 22 Benignbrpm WIIIOUR(22S.O) 0 BMnaitniptuunmt.~huni n.matuarneduunnrpya*rihitciedd(2h3n8)in I ~ u ~ u ~ ( ~ ~ I - I ~ ~ ,. E 1~1 Q I ~ I I ) 17.48 1-16 4-36 2l.M SMR 123-9 0 22.9 100-0 93% corgsdmrch l d -n.i-19~7 * 0.4-121.2 6S.4-l.wI.2 1 L I '.I An epidemiological study of petroleum rejnery employees IS is concerned in maintenance activities at any one that often found in cohorts demonstrating the time. Further,examination of specific crafts among "healthy worker effea." Comparison of the United the cases who worked in maintenance showed no States and the local county cancer statistics indicates unusual clustering. The most commonly listed craft that similar mults would have been obtained had wc was that of "still" or "tank" cleancr, but only five compared our cohort against county rates. We can individuals had this job title. Furthermore, three of xe no methodological flaw that might have prod@ the five held this job for three years or less. The this low SMR;we therefore conclude that this cohort lack of adequate historical exposure data did not has an unusually low mortality rate- Presumably, this pennit further analysis by quantification or even is a combination of the usual healthy worker effect qualification of exposure to specific chemical agents. (selection of the healthy for employment). a relative absence of d o u s risk from the job, and the effects of A N A L Y S I S FOR MALIGNANT, BENIGN, A N D continuing employment with its several benefits "N A T u R E uNS P ECI F IE D" E R AI N T U M O u R S (including health care), and lifestyle. The petrochemical industry has had considerable Because of the essentially negative findings, wc cal- interest in brain tumours and. specifically, brain can- culated the minimum relative risks that were 'cer, since the publication of several studies that detectable with 80% power at the 0-05 level for reported a raised risk.' Is-" Frequently, brain can- several causes of death of interest. For most malig- cers are incompletely artifkd and lack histological nant c a u s a of death that have been implicatcd in pn- diagnosis. To overcome this underreporting of malig- vious studies of refinery workers, the cohort as a nant brain tumours. and to recognix the lethality of whole was sufficiently large to detect with 80% power benign brain turnours, some investigators (NIOSH, any significant excess risk greater than twofold. In for example) have combined benign and nature particular, the minimum relative risk was 2.03 for unspecified brain tumours with brain cancer asa cate- brain caner, 1-47 for respiratory cancer. 1-70 for gory straddling three ICD divisions. stomach caner, 1-84 f m ~ a 1-7.9 for non- Table IO shows the analysis of all brain tumours Hodgkin's lymphomas, and 2.14 for kidney cancer. (classified as malignant, benign. or unspecified) The petroleum industry, especially in Texas, has among men. National age-race-sex-specific d y t h had considerable interest inbrain tumoun and brain rates for benign brain tumour (8th ICD 225.0) are not canar. To overcome the frequent underreporting of published but were obtained for this analysis by spe- malignant brain tumours, and to rrcognise the lethal- cial request to the National Center for Health Statis- ity of benign brain tumoun, we have combined r tics (NCHS). In calculating the expected deaths for benign and "nature unspecified"brain tumours with benign brain tumours the 1968 United States death brain cancer as a category straddling three ICD I rates were w d . Before 1968, no separate rates were divisions (table IO). Our analysis indicates that thcre gcneratcd for benign brain tumours. In addition. was a slight (non-significant) excess in malignant SMRs for brain tumours, nature unspecified (8th brain c a n a r @rimanly from Richmond) but a deficit ICD 239) are also presented in the table. Although in the other two categories, particularly io the cate- there was a slight non-significant excess in malignant gory of brain tumoun, nature unspecified. For the I cancer of the brain and central nervous system, three categories as a whole, the observed mortality i deficits were observed for benign and unspecified was exactly as expected. One possible explanation for brain tumoun. When all three categories were com- this observation is that diagnostic practice and dcath bined, the observed number of deaths (23) was identi- artification for brain cancer in Richmond and El cal with the expected (23.0). I Scgundo. both within large metropolitan arcas (San Franasco and Los Angeles. respectively). are more Dixlasioa specific and complete than the nation as a whole (19 of the 22 subjects with malignant brain canar lived in The total cohort was relatively large and was ade- the San FranciscO Bay area or LosAngela County at quate for examining mortality from several individual the time to their deaths). A previous study had indi- causes. Although the Richmond and El Scgundo cated that among the medically well insured, a high refineries had a few diflerencct in cause specific mor- level of investigation may lead to an increased fre- tality, their experience was sufficiently similar to jus- quency of the diagnosis of bnin cancer. giving an tify combining them . in the analysis in order to apparent (not real) increase in reported incidena incrcasc statistical powcr (in analysing lcss common rms."' Within thc Chcvron cohort, howcvcr, other UUSQ of dcath). explanations u n n o t be ruled out. Thc all causcs SMR was*only 72.4. a highly For malignancies of the lymphopoietic (lymphatic significant reduction from that expected on the basis and haematopoietic) system, there was a slight. non- of United States statistics. and somewhat lower than signifiant excess (the study had 80% power to exclude II I'- i i: i.16 Wmg. Morgan. Bailey. Swcncicki. Claxion.and KheiJets only an SMR of 150 or higher). Furthennore. there torits of the 37 deaths from this cause of death was was a trcnd of increasing SMRs with h t h increasing conducted. No unusual work assignment pattern was I duration of cmploymcnt and latcncy sincc first employment. This pattern was compatible with an exposure relatcd disease, although numbers were too small (within each time interval) to be conclusive of a dctcctcd uiid wc fiiild to idcntily m y common cxposure pattern. This observation is consistcnt with two recent casecontrol studies of lymphatic cancer among refinery workers" (Divine BJ. unpublished I II trcnd. Whcn the subutcgorics within all lymphopoictic u n c c r wcrc cxamincd. thc cxccss was round to come entirely from lymphosarcoma and rcticuiosarcoma. report). Based on similar methodological strategies, both concludcd that thcrc wcrc no unusual associations by work catcgory (proccss unit or job task in the case of the study by Divine). ! i ! as well as other lymphatic tissue cancers (non- Overall. the study has provided considerable reas- Hodgkin's lymphoma, multiple myeloma, and poly- surance concerning the long term heallh of Chevron cythacmia vera).There was a deficit in Hodgkin's dis- refinery employees. Total mortality was low, and ease (SMR = 52-1. not significant). but the number there was no suspicion of an excess risk either for lung was small. For leukaemia, there was also a slight cancer (an important issue in both Contra Costa and deficit (SMR = 88-3, not significant). As previously noted. Richmond and Ei Segundo refineries diKeral somewhat, with Richmond having a significant excess in other l-ym-~haticcancer for Los Angeies counties)or kidney and digestive cancer (prcviousIy reported in refinery workers)..A recent study suggests that in Contra Costa County, A d e n tial exposure to petroleum and chemical air emissions was associated with an increased incidence ofcancer 1 ! some difference in total lymphopoietic cancer urpcriencc, with Richmond showing an excess (SMR = 118, not of the buccal cavity and pharynx.2z In our study the SMR for this cancer was significantlylow for thc rata1 cohort (SMR = 47-0. p < 0.05). and for the Richmond refinery in Contra Costa County the SMR was 57-2 (not significant). For cancers of the lympho- poietic system. the study is moreequivocal. Although I1 significant) and El Segundo a deficit(SMR = 82, not significant). The difference between the two &nenes was not significant. Among the work location sukohorts, Lhm no major differences in mortality patterns. Maintenance workers had less of an excess oflymphopoietic cancer than did laboratory or operating personnel. The latter was the only cohort with an ex- ofleu- kaemia. We suaest caution in interpreting mortality pat- tcms, by diagnosis within the category of lymphatic no statistically significant increases in lymphopoietic cancer were observed, the results have raised quations concerning the potential risk to subcohorts or specific exposure categories. While the exocs~risk appears confined to non-Hodgkin's lymphatic cancer, not leukaemia. among employas at Richmond hired in or before 1948. and a detailed review of`the work histones of these casesdid not show any unusual dustering of work assignments. we could not NICout the possibility that exposure to a specificchemicalor mixture caused or contributed to the excess. Without a&- i I ii I and haematopoietic cancer (ICD 200-209). With the quate historical industrial hygiene data. however, an possible exception of Hodgkin's di- of recent exposure spCCificcasccontrol study of lymphatic candiagnosis, there is considerable opportunity for mis- cer is not possible. i diagnosis, misclassification, and even multiple diag- While it is safe to conclude that exposurc in general noses among the other rubria in the c a t q r y . For in the refineries. as well as exposures to many sptcific instance. both lymphosarcomas and myelo- chemical agents. have dmeased sizeably since the I i proliferative syndromes often end in acute myelocytic 1930s to mid-1950s (the time when most cascs were modleukaemias. Furthennore, it has bcm suggested that employed), it will be necessary to follow up the full most lymphatic and hacmatopoietic malignancy cohort for an additional to assess adequately j reflects an oncogenic insult to the pr- tan whether the risk of lymphatic cancer has declined in series." employees h i d after 1950. Because only first and last job were coded, it is difficult to draw firm conclusions w n m i n g the effect (if any) ofsubcohort membership on a particular disease category of interest. such as the lymphatic Rquetts for r.eprints to: Dr 0 Won& Esyifonmental HM- hc,520 Third Street -Suite 208, Oakland CA 94607. and haematopoietic cancer. The split into subcohorts is most.useful in examining general patterns only. In Refview of the excess in mortality from non-Hodgkin's lymphatic Caner a rCvicW Of the compkte work his- `Hanu NM. S18mky KM. Fowkr J L C a m moruliiy in oil vets I was I wils won with inccr ished ,gies, soci,k in reas- (ran and lung and ncer cent ten:om xer the otal tchwas CLIC la- or isk .cr. red wk us:he iX- lesn !n- Eil fic . he 're JlI 'IY in al 8. An epidemiological sf udv O/ petroleum refinery enqdoyees 17 refinery workm. J O c y .Urd 1979;21:167-74. rrcomnundaiionr of ilu eighih rrdsion confrreue. 1965. and IHunir NMTHnlmcr TM. Shdcntsrgcr LG. J o n a KE. Epi- Culoprrdh! rhr nmrirnrth Ii'wid I l r d h Assrmhly. V d r I and 11. h i o l o g i c study uT rdincry and chelnicd p h i workcn. Gncvill WIIO. 1967. IWX. JOrwp AfrJ IYX?;Z4203-I?. "Mursh GM. Prcininser M. OCMAP: 8 uur-onentd 0mrp.tioCul Rurhton L,AWcrson MR. An cpidcmiologial s u r y of eight oil cohort mortmlity rnrlyris program. Amrricun Sfaikiich ~fincriciln Britain Br Jlrul hfrd I9SI;M:225-34. P%l0945(:254. Rushton L Alderson MR. A cawcontrol study to invatigotc [he I' Rinsky RA. Young RJ, Smith AB. Leukcmi8 in bmzcnc worken. association h w c c n exporumto benzene and deaths from kukc- Am J Ind Mrd I981;2:2174J. . m u in oil rchmry workers. 8r JCuncer 1981:43:77-84. 'Schottcnfcld D. Warshaua ME.Zaubcr AG. Mcikk JG, Hart BR. I' Wong 0. An hdwiry-widc morialiiy rrudy of chemical worker3 occuparionaity cxpasrd i o bm~rne(.Technical repMc submitted A prospective study of morbidity and mortality in ptroleum to chemical Manufaaurar Assocktion. 1983.) industry employe~rin the United Statu - a prdimiary rrpon. "Theriault G. Gouk L A mortality study of oil refumy workm ICBanbuy v n No 9. Qwt$raiian of occupnihd CMCCI. JOCCUPM d 1979;21:367-70. -New Yo* Cold Spring Harbor Laboratory, 1981:247-65. l . A h u n ~V, Lcffingwdl s$ Uoyd JW. r i d . Brain in pa- 'Tabenhaw/Coopcr Asrodatu A moridiiy snu(~. of prrrolnvn refinery workers 1974. ~AmcricanPetroleum Institute Mediul rochemlcll work- a use-saiu report. Am 1 Ind Mrd 1980;1:115-23. Research Report No EA 7402.1 I' Wvaweilcr RI. Akxander V. Ixffingmll SS,r i d . Morutity from 'Thomas TL, W a w e i l a RI. Mourc-Enlo R, b y 8 S. Fnumeni JF. brain tumor and otherauyfin a cohort study of pccrochcmiul Mortality patterns among w o r k m in three Texu oil refincriu JOcmp Mrd 198E24:135-41. 'lsai SP,Wen CP.W&SSNS. Wong 0.McCldian WA. Gibson R L workerr JNCI 1983;7&7SSI. -"Grccnwaid P. Friedlander ER. Iamncc CE.H e a m T.Euk K Diagnostic sensitivity bias an epidemiologic explanation l i Retrospctive monalily and medicrl surveillance studia of an apparent b n i n tumor excess JOcorp Afed 1981;23:690-4. workers in benzene arcas of rdincria. J Ocmp Afed 1983; I' Bakhrhi A. Minowada J. Amold A. ciul. Lymphoid bku crises of 29685-92. chronic myelogenous leukemia repracnc s q a in Ihe devd- 'Wen CP, TYi SP. Wciu NS. McCteUan WA. Gibson R L A opment of BCcn praurson. N &gl J M e d 1983W826-31. population-based cohort s t d y of b n i n tumour monality lo K ~ r JqH. chronic myelocytic (multipotcnt-sicmcdl) kukcmia. among oil refincry worken with a discussion of methodological N J M d 1983~3WBSI-2. issues of SMR and PMR. In: Banbury rrporc No 9. "Thomls fl, w8XuKikf a. Cnndall MS. While DW. Mourn Quani@cuiion of OrrUWiiod rawer. &VI York Cold Spring Enso R F n m iJF. Goccr morulity patterns by wo& ate- **Habour Labontoy. 1981:413-32. loWen CP.Tsai SP, McClclhn WA. Gibson R L Long-(- mor- goy in thra Texas oil re6nCria. Am J l n d dled 1984;6:>16. Kaldor J. Hank JA. G k Er i d . S u d s r i a l association between tality study of oil refinery worken. 1. Mortality of hourly and aam incidence and major cause of monaiity, and arimatcd d u i e d workers. Am J Epidcmiol 198k1l 8 5 2 6 4 L rrridcnlial uporure IO air emiuions from pcirolcum and c h m - 'IWorld Health Organintion. McUruLII uf the inrrmaiionol rrarhckal iCrl plants. firLon ffedrh Prrrp 1984:U.319-32. c)prri/icoiion of dkasrs. ujuricland cmucs of &ah. h r d m ik I