Document wgb4vVzLn40ODNk4e31wXYzEo

Mortality among Workers Exposed to Coal Tar Pitch Volatiles and Welding Emissions: An Exercise in Epidemiologic Triage ~ ~ I C H A ESLI L V E R S T E I NM. D . MPH. N EIL MAIZLISH,P H D . ROBERT PARK, MS, A N D F RANKLIN M I R E R .PHD /- Abstract: The United Automobile Workers International Union ,,tablished a system of epidemiologic triage to evaluate patterns j?mJjonality among groups of union members. In response to worker ,30cem~t.he Union examined mortality at a metal stamping plant. ,,toe 3 method which linked pension records with the State of computerized death registry. The observed proportion of 4ignant neoplasms was nearly twice that expected (95% Confiynce Limits 1.36. 2.621. Two- to five-fold excess proportional #dality from cancer of the digestive organs. lung cancer. and edkrmia accounted for most of the overall excess. Strong associaNns *ere found between lung and digestive organ cancer and employment as maintenance weldersor millwrights in the plant (odds ratios > IO). High levels of six polycyclic aromatic hydrocarbons with mutagenic and carcinogenic properties were found during hot coal tar application to wood block floors. work conducted by the high-risk groups. These levels were substantially reduced following the purchase of new tar pots. The example demonstrates that epidemiologic tools can play a valuable role in occupational health decision making, but care must be taken to avoid mechanical reliance on quantitative testing and to acknowledge the important role of social and political value judgments in the establishment of responsible public policy. (Am I Public Healrh 1985: 75:1283-1287.) / ~,,,odlrL'fion 1 This paper describes recent union-based efforts to de- ,,<lop and use epidemiologic surveillance techniques as utin ne. accessible tools in the day-to-day conduct of a health i' ,,d safety program. The United Automobile Workers International Union :c,dw)represents I" workers in more than 2.500 rorkplacesthroughout the United States and Canada. These &rrs face an enormous range of potential health hazards jobs ranging from paint formulation to salt mining. iutornobile manufacture, employing one-half of UAW mem- kn,is itself a collection of many distinctly different types of bork. The UAW has had an active health and safety program br many years, utilizing occupational health professionals d local union health and safety representatives to identify workplace hazards and to develop technical and political mtegies for their control. Stimulated by recent intense membership concern with apparent excess cancer mortality ovanous workplaces. it became apparent that the health and dety program needed a new dimension-ne which could mJuate patterns of illness, injury, and death that might not k apparent to even a skilled observer with the training to +cognize known hazardous materials and conditions. A practical system of epidemiologic triage was required w that threshold decisions about health and safety strategy auld be made on firm grounds. For example. with several iorkplaces reporting apparent cluster5 of cancer deaths, a :rompt choice among three courses of action is necessary: 1) ipnority program to reduce worker exposures to known or :kdy hazards through engineering or other controls: 2 ) .:5ternatic research to explore suspected health effects wlemented with modest. interim protective measures: 3) XNIKlnCe t0 workers that work-related health problems not been demonstrated and are to have OC- .umd and that little additional action is needed. It is not From the L'nited Automobile Workers. Health and Safety Department. urers repnnt requests lo Michael Silverstein. MD. MPH. C A W Health and h Department. 8000 E. Jefferson. Detroit. M I 48214. Dr. Majzfish is mntlv uith the Occuaational and Environmental Health Unit. Ilnivemitv of reasonable to expect that any epidemiologic method can determine unequivocally which of these activities is correct for a given workplace. Moreover, many epidemiologic tech- niques are too slow and cumbersome to have short-term decision-making value in the workplace. However, it has been possible to adapt standard epidemiologic methods to Serve as public health tools in the so that quantitative data can be used in among alternatives at a given workplace and in determining priorities among different locations according to relative urgency. One such method, addressing mortality, has been established in Mich- igan by making use of the State's computerized death registry. Three steps are involved: First, when a potential problem is identified. Local Union and international Union records are searched to produce a listing of persons at risk, or known to have died, with names and social security numbers.The full cohort of interest, all those who worked in the plant over a defined period, is usually not accessible or cannot be generated quickly. However, an inventory of available records (e.g.. seniority lists, dues cards, application records, and some- times death records) usually provides a fairly complete specification of those at risk. The fact that such a list might have gaps and sampling inconsistencies is not a deterrent in this triage process, as long as there is reasonable certainty that major confounding and selection bias have been avoided. This can often be assured by using a subgroup for which nearly complete and unbiased ascertainment is possible. such as all those hourly employees with at leas! 10 years of credited pension service. Second. names and rocial security numbers on the list of deceased or at-risk persons are matched against comput- erized files such as [hose obtained from [he State of Michigan ~~~~~~~~t of public Health. The Michigan files contain all in-c,tate deaths since 1970 with underlying cause coded by the Classificationof Disease (ICD) classification revision in effect at the year of death. As a useful but not essential adjunct niit-nf-qtnte deaths are identified by searchc- -P (SSA) records. \ 1985 .American Journal of Public Health 0 0 3 6 / 8 S S 1 . 5 0 '.-T'lD .-... I ...... .-..-.-..-, 0-..-.--- ---..--.----- proportional mortality ratios (SPMR). lp'dNovember 1985,Vol. 75,No.11 1283 : SILVERSTEIN. ET AL. I As an example, we describe the course of a mortality reports of deaths or copies of death certificates of investigation at a metal stamping plant where maintenance workers and retirees. Additionally, the local union new welders and millwrights were exposed to coal tar and welding circulation file was used to identify those who left e emissions. ment before vesting or left employment with vested pen Methods rights but might have died before retirement. This file matched to SSA records of deceased persons using S Development of Study Security number as the match key. The study site was a metal stamping plant located near Kalamazoo. Michigan. The plant. which began operations in 1966. has employed on average approximately 3.000 UAW members. The plant operations have included metal stamping, spot welding. tool and die making. metal machining, and pattern making. The Local Union became concerned in 1980 about a possible high number of cancer deaths among its members. The Local Union Health and Safety Representative began to compile a list of deceased members along with information about their causes of death and jobs. which he was able to abstract from information available at the plant. The U A W had not yet developed most of the epidemiologic tools described in this paper. However, manual calculations of proportional mortality were done and did not reveal excesses of cause-specific mortality or associations between mortality and specific work assignments. By 1982, the number of known decedents was 58 and the Ninety-eight (98)deaths were reponed from all sou including eight uniquely ascertained from SSA records judgment is that the group of 98 decedents is a virt complete ascertainment of hourly workers who died w active or retired: a small number ofdeaths among worker deferred vested status. or those terminated with less than years seniority may have been missed. Death certificates (or. in two cases. abstracted recorc: were obtained for 94 of the 98 deaths (96 per cent). For 611r the Michigan deaths. underlying cause of death was detc: mined by record-linkage of the death file to the computeriztc Michigan death certificate registry using numbers for matching. The remaining deaths. which includa out-of-state deaths and those which occurred before the S m computerized its files in 1970. were coded according to lbt 8th Revision of the ICD by the investigators. The final study group included 91 White males. 2 Blad males. and 1 White female. 'Union had developed the computerized capacity for PMR analysis. Twenty-two cancer deaths among White men were observed, compared with eight expected based upon national rates. The Union decided to continue the investigation on a high priority basis. A site visit was conducted by International Union staff to gain insight into the nature of chemical exposures potentially associated with various jobs. It was observed that the floors in this plant consist of wood blocks which are set and protected with hot coal tar in an operation that closely resembles roof tarring. Further, employees reported that: I ) millwrights have been responsible for laying the floor tar and have been exposed to smoke and vapors from this work; 2) the millwrights at this plant also work closely with the maintenance welders who weld or flame cut floor-bolts at Work Histories Personnel records were obtained from the employer f@ 83 of the study group. Following site visits, discussions @ the plant management and the Local Union. and a review 11 the available industrial hygiene reports. decedents grouped into two exposure categories: I ) maintenance ers and millwrights; and 2) all other job classifications. 'Tk+ classification scheme was based on the judgment that mab tenance welders and millwrights are most likely to be posed to coal tar pitch volatiles. Pre-plant employment as a welder or millwright sa coded on an ever-never basis. Statistical Analyses SPMRs. standardized for age, sex, race, and calen' floor level where tar and wood blocks sometimes ignite and generate smoky emissions; and 3) the wood blocks have been treated with wood preservatives, and have absorbed solvent and oil spills. Historically. maintenance welding and floor tarring work had been done with a minimum of local exhaust ventilation or respiratory protection. Routine quantitative measurements of these exposures had not been made, but in 1979 personal and area air samples taken by the employer during floor tamng revealed coal tar pitch volatile concentrations in excess of 900 ugkubic meter T W A (8-hour time weighted average) with peak exposures much higher. This additional information about the chemical environment prompted a decision to expand the mortality evaluation and. simultaneously, to press for reduced exposure to tar emissions. The outcome of this more thorough review is presented below. Study Group The target population for the investigation was all deceased local union members ever-employed at the stamping plant with a date of death between January 1. 1966 and December 31, 1982. Sources of data included local union lists of deceased workers. union pension records. and employer life insurance records. These sources routinely received year of death. and case-control studies within the decedc9 population were carried out. Standardized Proportional Mw lalit). Ratios (SPMR) for 54 causes of death were calculald for the 91 White males with U S White males as the referrm' population. updated to include National Center for Healft Statistics (NCHS) mortality data through 1978. SP3fRs war also calculated for restricted subgroups based on em$' ment duration and prior occupational history as a welder millwright. Case-control analyses were designed relationship between cancer mortality and spe share. among other things. hot tar operations flame cutting near tarred surfaces. Cancer ca by underlying cause of death and groupe specific ICD codes: lung, 162: digestive organs. leukemia. 204-207: and all malignant neoplasms: Controls were defined as all non-cancer deat nant respiratory disease deaths were therefo controls in the case-control analysis of lune sion made to conserve statistical power despite the PoSaF,lrrr that shared etiologic factors might result in a reduced to demonstrate associations. "Exposed" "3s dcfinCJ ever-maintenance welder or ever-millwright anJ posed" were all other job classifications. s3Cases and controls were drawn from the jCc 1284 AJPH November 1985- ,flL~ 1-Standardlzd Proportional Wrtallty, Whltm MaW 1966-82. by C a u r oi Doath /--- Cause of Death I Observed Expected SPMR 95%C.L. 28 1482 189 1.36. 2.62 -2 0.41 485 4 0.94 428 1.75, 10.45 22 26.79 82 0.59. 1.13 MORTALITY AMONG STAMPING PLANT WORKERS: TRIAGE TABLE Shaoclstlon bdwoon All C . n c ~m i l t y md w.ldw;ly, mlpht Classlflcatlon Classification All Ages CaSeS cultrob O-R_---10-.4-;9_5%.C_L: 2_.64_2.2. - . _ _ _ _ _ _ . _ _ _ _ _ _ . _ _ . . ~ 4 Y0ears .-- 2-Charact.r(dlcs of Cams and Controls L, Charactenstics Cases n = 27 rp at Death. Mean 60) at Death. Range in. Mean (SO) ..Y.l_._..... Jwnghts. Number (Oh) 47 (1 1) 22-66 9 (5) 7 (26) controls n=56 38 (12) 2048 7 (4) 4 (7) bithcomplete work histories: 80 White males, 2 Black males. ad I White female. Tobacco smoking patterns were not &vn. Stratificationwas used to control potential confound@ by age and pre-plant occupation. Iesults The observed number of malignant neoplasms (Table 1) inearly twice that expected (PMR = 28/14.82 = 189, 95% :.L.1.36, 2.62). Two-fold to five-fold excess proportional wrtality from cancer of the digestive organs. lung cancer, Ipd leukemia accounts for most of the overall cancer excess. jsentially the same excesses were found in the subgroups tith five or more years seniority or with no pre-plant Imployment as a welder or millwright (data not shown). Table 2 gives age and occupational characteristics of mes and controls. O n averane. hnrh cancer cases and ontrols died at relatively young ages and short employment uration. Because cases tended to be older at death than mtrols and were overrepresented by former welders or mllwrights, case-confrol analyses were stratified by age and xe-plant occupation. Table 3 presents age-stratified odds ratios for the asso.;ation between cancer mortality and employment as main'tnance welders and millwrinhts. The mean aee in the 40+ stratum was 50.3 (49.4 among nine welder/millwrights md 50.6 among 34 non-welder/millwrights)and in the <40 qe stratum was 30.9 (33.8among four welder/millwrightsand 3.6among 36 non-welderimillwrights). Odds ratios of more baa 10 were found overall and in the 40+ age strata. jimilarly. the odds ratios for lung cancer and digestive organ a c e r s were highly elevated in the 40+ strata (Table 4). Table 5 presents cancer mortality and welder or millaright classification stratifying on prior welding or millwright stcupation. While odds ratios are elevated in both strata, the aids ratio was much greater and statistically more stable in subgroup without prior welding or millwright occupa- J@ns.Odds ratios for lung cancer or cancer of digestive *Bans followed a similar pattern. Further. case-control analysis simultaneouslystratifying [UPH November 1985.Vol. 75,No.11 40+ Years Wekler/Milkrght mer OR = 14 7;95% CL: 1.6-131.6 8 12 1 22 TABLE 4-Assoclatlon botwoon Csncor Mort.llty, Two Sltos, and WeldwlMlllwrlgM ChMlfiCatlOfl, AO, 4 O t Y u n at M h classification Lung Cancer CeseS Gnirds Welder/Millwright 3 Other 5 O- -R-_-=__.13.-2-:-9- -5.% C__L- _:_1_-.1.-1-5-4-.9- - - - . . - - - - - - - - - All Oistivs Cancers 1 22 Welder/Millwnght other OR = 16.5;95% CL: 1.1-162.2 3 4 1 22 TABLE cAuotlrtlon ktmrnA"--lw tlon by Prlor occuprtlan and Job Classification Former Welders or Millwrights cases controls We!dw/MilIwnght 6 2 Other 12 O-F-I =-'-.O:-9-50.'0-cL-'0-.3--1.07-.0- . - . - _ _ . _ _ _ _ . _ _ _ _ _ _ _ _ Not Former Welders or Millwrights WeldecMillwright Other OR = 12.8;95% CL: 13-122.5 - 4 16 1 51 on age. employment duration, and prior occupation also revealed similar associations between cancer mortality and millwright or maintenance welder classification, but statistical power was diminished, as expected with decreasing cell sizes. Discussion A union-based epidemiologic surveillance program proved to be an effective and efficient guide to occupational health action in a metal stamping plant. Use of a mortality surveillance system designed for routine use revealed excess proportional mortality from cancer of the lung, digestive organs. and leukemia. This lent credibility to local union concerns and led the union to assign this situation very high 1285 SILVERSTEIN. ET AL. TABLE 6-Stamplng Plant Exposum DeW Before and A t t r Installation of New Tar Pot# Chemical Old Tar Pot: Vent OM Tar Pot: Floor New Tar Pot: Fbof Phenanthrene 500 50 0.88 Anthracene 200 20 0.55 Acndine 10 1 ND" Pyrene 100 10 0.16 Chrysene 20 1 ND Benz(a)pyrene 1 ND ND 'All values in mgcu m Shor(4erm Air samples "Nm-DetectaMe priority for further investigation and to undertake interim efforts to control suspect chemical exposures. Subsequent case-control analysis documented associations between mortality from lung cancer and digestive cancers and work as maintenance welders and millwrights which involved probable exposure to coal tar pitch volatiles and welding emissions. This served as the basis for a further decision to seek more vigorous in-plant environmental intervention. Following discussions of the findings the employer took short-term source and area samples during tarring which showed substantial levels of six specific polycyclic aromatic hydrocarbons with known carcinogenic or mutagenic prop- erties (Table 6). Agreement was reached to take steps to reduce exposures of maintenance welders and millwrights to the suspect emissions. The employer purchased new, thermostatically controlled electric tar pots and agreed to obtain more than 20 portable ventilation units. construct several exhausted welding booths, purchase additional respirators, and initiate a search for non-coal tar floor coating substitutes. A series of educational meetings was held with more than 200 affected workers in the plant. At the urging of the union, the employer sent a bulletin to its other locations asking that floor tarring operations be evaluated and that steps be taken either to reduce emissions or find substitute materials. Follow-up area air sampling during tarring was conducted after the installation of the new tar pots and showed large reductions in airborne polycyclic aromatic hydrocarbons (Table 6). We believe that these findings are sufficiently plausible to warrant an alert to the public health community of a probable. but previously unrecognized, hazard. The observed associations are large: feasible control measures are readily available: and the findings have considerable biologic plausibility.24 Exposure of roofers and coke oven workers to coal tar pitch volatiles containing polycylic aromatic hydrocarbons has been associated with excess multiple-site cancer mortality.5.6 The group experiencing excess mortality in this study uses similar materials under analogous condition^,^ and exposures to specific mutagens and carcinogens have been documented (Table 6). Efforts should be made to replicate the findings at other locations where similar work is done in such a manner that the combined and independent effects of coal tar and welding exposures can be evaluated. In our judgment, concurrent efforts should be made to notify representatives in potentially affected workplaces of the possible hazards and to encourage actions to reduce worker exposures to chemical emissions during floor tarring operations as the prudent response to the ~urveillancefindings. Mortality surveillance enabled the union to distinguish this plant situation from others which. on the surface, might 1286 iolOgic science in workplace health and safety prowe saution against the tendency to substitute numbers pproach in this regard as before us, not what we MORTALITY AMONG STAMPING PLANT WORKERS: TRIAGE ACKNOWLEDGMENTS This work would not have been done without the effons and determination of Jack Kelly and Roger Snell of UAW Local Union 488. Kalamuoo. Michigan. This research was supported, in part. by NIOSH Contract +?10-81-5104. Some of these data were originally presented at the American Public Health Association Annual Meeting, Dallas, November 7. 1983. eports before deciding to warn workers of and before taking protective actions, we 3 carcinogen freedom to roam through the plant during thy trial is far more dangerous than letting an accused free on bail."* D: Problems and challenges facing occupational medicine in the from labor's viewpoint. Presented at American Occupational nference. Detroit. MI. U2Y80. REFERENCES I . Monson R: Analysisolrelative survival and mportional monality. Comput Biomed Res 1974: 7:325-332. 2. Lindstedt G. Sollengerg J: Polycyclic aromatic hydrocarbons in the occu- pational environment. Scand J Work Environ Health 1982: 8:l-19. 3. Bingham E.Trosset R. Warshawsky 0:Carcinogenic potential of pctmkum hydiocarbons-a critical review. J Environ Pathol Toxicoi 1m:33483-563. 4. Cmke M. Dennis A. Fisher G (edsJ: Polyarornatic Hydrocarbons. 6th International Symposium. Columbus. Ohio: Baitelle Rcss. 1982. 5 . Redmond C. er ul: Long-term mortality study of steelworkers: VI.mortality from malignant neoplasms among coke oven workers. JOM 1972: 14:SMI. 6. Hammond C.et a/: Inhalation of benzypyrene in man. Ann NY Acad Sci 1976: 271:116-1~4. 7. Dinman E. Sussman N: Uncertainty. risk and the role of epidemiology in public policy development. JOM 1983: 2551 1-516. 8. Lanes S. Poole C: Truth in packaging?: the unwrapping oiepidemiologic research. JOM 1984: 26371-574. .has bee :rs and al 3, tiicohnSt h. eU A New Publication: The Journal of Rural Health I 2 judgrnen bred by t cal conce :d are in lude that 3er cent C ent about w h is rno t decisions J U t O f 'I The Journal of Rural Health. a new periodical which began publication in January 1985, Semes as a medium for communication among health scientists and professionals in practice, education, and research settings. The principal goal of the journal is to advance professional practice, research, theory development. and public policy related to rural health. According to its sponsoring organizations-the American Rural Health Association, and the National Rural Health Care Association-"For some time there has been a cadre of rural health professionals who have not had an adequate communication channel for their views, ideas, activities, and scholarly research results. The Journal of Ruraf Healrh will fill that void and provide further identity to the field of rural health. This identity is essential if the world around us is to become sensitized to the vast differences which exist between rural and urban America. and to evaluate the health care consequences of those differences." The peer-reviewedjournal will be organized around six sections, to appear routinely in the journal when a sufficientnumber of manuscripts have been reviewed and accepted for publication to produce a meaningful contribution to the literature. The six sections will be: 0 Education for Rural Health Professionals 0 Rural Health Policy 0 Rural Health Practice Applications 0 Current Research in Rural Health 0 Book Reviews 0 Research Reviews Contributed manuscripts are welcome and should be sent directly to the journal editor, Raymond T. Coward, PhD, Center for Rural Studies, University of Vermont, 448 Waterman. Burlington, VT 05405-0160. Instructions for manuscript submission appeared in the January 1985 issue of The Journal of Rural Healrh. or can be obtained by writing to the editor. Subscription information is also available from the editor. Book Reviews-To be considered as a potential book reviewer, to suggest books that might be critiqued, or to reply to a particular review, write to: Dr. Ben F. Banahan, 111. Health Services Research Division, Research Institute of Pharmaceutical Sciences, University of Mississippi, Universi- ty, MS 38677. 1287 If t Mortality Among Workers Employed in Petroleum Refining and Petrochemical Plants T u r y L Thomas, M.S.; Pierre Decoufle, Sc.D.; and Rafael Moure-Eraso, M.S. The cause-specific mortality experience o i 3,705 members of the Oil, Chemical and Atomic Workers International Union was examined to determine if there were unusualpatterns of fatai disease that may be indicative of h k u s agents in the work environment. Deaths among a c e Union members that were reported by locals in Texas between 1947 and 1977 were identified through membership records, and proportionate mortality was d M l y z e d in several broad industrial categories. PMRs for W X ~ Sof the !her and D i l k r y Passages, pancreas. lung and skin were elevated among refinery and petrochemical plant workers; however, risks did not increase with l e n s h of membership. Increased relative frequencies of stomach cancer, cancer of the brain, leukemia and multiple myeloma were confined to white males in the same UteSor~who had been Union members for 70 or more Years. Excess deaths from stomach cancer and brain cancer found among white male members employed at one SPecific oil reiinery and petrochemical plant. Observed of deaths from cancer o i the stomach were W a t e r than expectedamong whites and nonwhites, and an elmated PMR for lung cancer among nonwhites was !ound at an additional plant. findings suggest that workers in this industry may be at increasedrisk of certain cancers and indicate areas for further investigation. several studies have indicated the need for further "ahation of cancer risks among persons emploved in Petroleum refineries and petrochemical plants In a recent study of cancer mortalrtv bv county. high rates of cancers o i the lung, nasal cavity and sinuses, and skin were found among male residents of counties where the petroleum industry is heavily concentrated.' Early studies indicated increased risk of scrotal cancer among wax pressmen in a refineryzand cancer of the larynx and nasal sinuses among workers engaged in the manufacture of isopropyl alcohol.' Although no increased cancer risks have been found in more recent studies ot US. reiinery workers,' a Canadian study has reported elevated risks of esophageal, stomach, and lung cancer among workers exposed to crude petroleum and its productr.' The opportunity to evaluate cancer risks among a large group of workers employed in petroleum refineries, petrochemical plants and related industries arose when officials of the Oil, Chemical, and Atomic Workers International Union (OCAW) invited the National Cancer Institute to review data sources within the Union that might be useful for a epidemiologic study. The present report examines the cause-specific mortality experience of Union members to determine if there was an unusual pattern indicative of hazardous agents in the work envirocment. The purpose ot this study is to point out areas for further investigation. Materials and Methods In considering what kind of study design might be possible, a retrospective cohort study was deemed not feasible because of limitations in the Union's recordkeeping system. Since records of deaths among active Union members are maintained by the Membership Department, we decided to examine the relative frequency of specific causes of death as a preliminary means of obtaining an estimate or site-specific cancer risks in various broad industrial categories within the petroleum refining, petrochemical, and related industries. Because ot the heavy concentration of the petroleum industry in the state of Texas, and the large OCAW 97 -Table 6. Mortality for Selected Cancer Sites in White Male Texas OCAW Members in Category 2 site (8th Revision, ICD) by Length of Union Membership. Y -I 2 Lenqth ot Membership 2 < 10 Years 1 10+ Years (N = 578) (N = 1144) 1 $ 0' E t PMR E t PCMR E TPMR 0' E t PMR E$ PCMR E TPMR Liver and biliary aassages (155 156) Pancreas (1571 Lung (162. 163) Skin (172 173) b 5 2 4 208 2 6 190 2 7 184 7 4 6 152 6 8 103 5: :22 15 4 4 3 3 8 7 5 3 2 8 1 7 5 2 2 3 9 1 '4 1 2 5 1 1 2 1 6 7 0 8 4 1 2 4 1 1 3 .! 35 21 7 1 6 1 1 24 3 1 441 25 9 1 35 95 74 3 1 281 91 1 1 0 4 76 6 1 241 1 16 2 3 2 5 6 1 2 7 2 2 3 3 4 1 7 9 8 4 9 1 6 5 6 2 1 2 9 6 3 1 2 6 '0 = Observed deaths t E = Expected deaths Sased on proponionate mortality for U S white males $E = Expecled deaths Oased on proportionate cancer mortality for U S white males 6E = Exoecled deaths based on Drooonionate mortality for Texas white males istatistically significant at 'he o 05 level relative frequencies of specific causes or death, excesses for one or more causes will automatically force others to be in deficit In addition, whenever multiple comparisons are made, a certain number or differences will be statistically significant based on chance alone Despite its limitations, a P M R analysis is usetul as a preliminary, hypothesis-generating step Results Distribution of study subjects by industrial category is shown in Table 1. ,Most of the decedents were classified into Category 2. Only 55 decedents worked for com- panies that could not be classified. Since the study group is comprised entirely of workers who were active Union members at the time of death, less than 1% were age 65 or older; however, over half of the 3.105 decedents were 50 or older (Table 2). Table 3 indicates that 62% of the whites and 57% of the nonwhites died 10 or more years atter initiation into the Union. A substantial number 0 25%) were members for 20 or more years prior to death. PMRs for major causes of death are shown in Table 4, by race, for each industrial category that had at least 100 deaths. Among whites, the relative frequency ot cancer deaths is elevated only in Category 2 . Ratios for heart disease are elevated in both racial groups and in each industrial category. Obsewed numbers of deaths from respiratory and digestive disease are lower than expected in each category and exceptionally low among whites in Category 4. An increased relative frequency ot motor vehicle accidents is seen among whites in Categories 1 and 3 and among nonwhites. Non-motor vehicle accidental deaths are elevated in each category except 3. Site-specific cancer m o r t a l i v is shown in Table 5 for each of the industrial categories in which there were at least 100 total deaths. Most interesting are results for the petroleum refinery and petrochemical plant category (Category 21. Among whites, observed numbers of deaths are significantly greater than expected for cancers of the digestive and respiratory systems, skin, and brain. A nonsignificant excess o f respiratory cancer deaths appears in the crude oil and gas extraction category (Category 1). No other site/category combination is associated with unusual mortality, although the numbers of deaths are small in many instances Observed numbers of deaths from digestive and respiratory cancer are greater than ex- pected for nonwhites. Further analyses of cancer risks i limited t o decedents employed in petroleum refined and petrochemical plants (Category 2). If occupational exposure contributes to cancer risk, generally expect risk to increase with increasing dura& of exposure, or, in this case, with length ot members Specific cancer sites that were associated with i n c r e 4 relative frequencies in Categorv 2 were examined 4respect to length of membership. Not all cancer sites elevated PMRs in the petroleum refinery petrochemical categov show a positive gradient. Table shows cancer sites for which risk among members in & category does not increase with length o i membershi The elevated PMR tor cancer of the pancreas is coniine to members initiated less than 10 years prior to deatj 14Observed numbers of deaths from cancers ot the 7lung, and skin are greater than expected regardless c 4length of time between initiation and death. PCURs puted tor each ot these sites show essentiallv the sanj <pattern for the 10 year membership group. howevd ratios are smaller for decedents who were members for! or more years. Results using Texas white males as a cotj parison group vary little from the expected values basj on the total U.S. males for each site except skin TPA tor skin cancer are much lower than P M R s but rerr elevated. 3 Table 7 shows cancer sites for which risk increases wil length of membership for white males in :he petroieu refinery and petrochemical plant categorv ,An inciease relative frequency of stomach cancer is conrin members who joined the Union 20 or more 'rears death. There were tour deaths trom multiple mcel the same length-of-membership group and onlv 1 expected. Deaths trom leukemia and trom c a n c m o brain and central nervous svstem seem to cluster am males who were Union members for 10 or more Although an increased relative frequencv or k cancer deaths is not present in the total group, there significant excess of deaths among members who w itiated at least 20 years prior to death, with seven observed and 3.3 expected. Expected numbers of based on proportionate cancer mortality are high those based on proportionate mortality, however. same trends are present. Table 8 shows cancer sites contributing to the exc shown in Table 5 for nonwhite males. Increased re1 100 Mortality Among Workers in Petroleum Refining and Petrochemical PlanWThomas. Decoufle. and Moure-E - -w dd F 5 N - 0P 0- O0W.- 5 W N 0)WW - A- P N W - -Na *3 - 0 uI I lwvl VlD -y N u bA7 4 'd W I A w A A- -A?mwU -0 l o~mb , N 0 A OD- - -W N 0 o ) U 0N mW0 -0 I I I IOU CON ww U 4 U NWP -7 N N W OAV w P w yl~m -a m NO NUO -a m o w- N- - -a0 o m A W U 00 -Oc Wo Wa mW w +P + 'm4 * mC FmW -wU P N mo -n Journal of Occupational Medicine/Vol. 22,No. 2/February 1980 99 site (8th Revision. ICD) Liver ana Diliary passages (155. 156) Pancreas (157) Lung (162 163) Skin (172 173) < 10 Ywrs (N = 578) PMR E$ PCMR E TPMR 0' , (N = 1144) '$ E t PMR Et PCMR E TPMR :. -. 5 2 4 208 2 6 190 2 7 i a 4 i 4 6 152 6 3 103 5 7 122 1 5 4 4 3381 5 3 2 3 1 1 5 2 2 a s i 14 12 5 1 12 16 7 0 34 12 4 1 13 35 21 7 1 6 1 1 24 3 1 4 4 1 25 3 1 35 95 74 3 1 281 91 1 I 04 76 a i 247 6 2 3 2561 2 7 223 3 1 179 8 4 9 165 6 2 129 6 3 ' 2 6 '0 = Observed deaIhs t E = Expected deaths based on proportionate mortality for U S white males :E = Expecied deaths based on proportionate cancer mortality for U S Nhite males 6E = ExDected deaths based on proportionate mortality for Texas white males iStatistically significant at the 0.05 level relative frequencies ot specific causes of death, excesses pected for nonwhites. Further analyses of cancer risks are for one or more causes will automatically force others to limited to decedents employed in petroleum refineris be in deticit. In addition, whenever multiple comparisons and petrochemical plants (Category 2). are made, a certain number of differences will be If occupational exposure contributes to cancer risk, we statistically significant based on chance alone. Despite its generally expect risk to increase with increasing duration limitations. a PMR analysis is usetul as a preliminary, ot exposure, or, in this case, with length of membership. hypothesis-generating step. Specific cancer sites that were associated with increased relative frequencies in Category 2 were examined with Results respect to length oi membership. Not all cancer sites with Distribution of study subjects by industrial category is elevated PMRs in the petroleum retinerv and ;hz:vn in Table 1. Most of the decedents were classified petrochemical category show a positive gradient. Table 6 into Category 2. Only 55 decedents worked for com- shows cancer sites for which risk among members ir! this panies that could not be classified. categorv does not increase with length of membership. , Since the study group is comprised entirely of workers The elevated PMR ior cancer ot the pancreas is confined . who were active Union members at the time of death, less t o members initiated less than 10 years prior to death. than 1 ?6 were age 65 or older; however, over half of the Observed numbers of deaths from cancers o i the liver, , 3,105 decedents were 50 or older CTable 2). lung. and skin are greater than expected regardless of Table 3 indicates that 62% of the whites and 57% of length of time between initiation and death. PCMRs corn the nonwhites died 10 or more years after initiation into puted for each ot these sites show essentially the same <the Union. .4 substantial number 0 25%) were members pattern for the 10 year membership group. however, for 20 or more years prior to death. ratios are smaller for decedents whowere members for 10 PMRs for major causes of death are shown in Table 4, or more years. Results using Texas white males as a c o n by race, for each industrial category that had at least 100 parison group vary little from the expected : ' j ! ~ nba~sed deaths. .\mong whites, the relative frequency of cancer on the total U.S. males tor each site except skin. T P M deaths is elevated only in Category 2. Ratios for heart for skin cancer are much lower than PMRs but remai disease are elevated in both racial groups and in each i n elevated. dustrial category. Observed numbers of deaths from Table 7 shows cancer sites for which risk increases w respiratory and digestive disease are lower than expected length ot membership tor white males in :he petroleum in each category and exceptionally low among whites in rerinery and petrochemical plant categoq ,An increased Category 4. An increased relative frequency of motor relative frequency ot stomach cancer is coniined to vehicle accidents is seen among whites in Categories 1 members who joined the Union 20 or more vears prior to and 3 and among nonwhites. Non-motor vehicle acciden- death. There were four deaths from multiple rnveioma in tal deaths are elevated in each category except 3. Site-specific cancer mortality is shown in Table 5 tor each ot the industrial categories in which there were at least 100 total deaths. ,%lostinteresting are results for the the same length-of-membership group and onlv 1 3 were .ol; expected. Deaths from leukemia and trom cancers of the .... ,+.brain and central nervous system seem :o cluster among > . males who were Union members tor 10 or more years. petroleum refinery and petrochemical plant category Although an increased relative frequency 01 k i d v (Category 2). Among whites, observed numbers of deaths cancer deaths is not present in the total group, there is a are significantly greater than expected for cancers of the significant excess of deaths among members who were irr digestive and respiratory systems. skin, and brain. A non- mated at least 20 years prior to death, with seven deaths significant excess of respiratory cancer deaths appears in observed and 3.3 expected. Expected numbers ot deaths the crude oil and gas extraction category (Categow 1).NO based on proportionate cancer mortaliw are higher than other siteicategory combination is associated with those based on proportionate mortality, however. the unusual mortality, although the numbers of deaths are same trends are present. small in many instances. Observed numbers of deaths Table 8 shows cancer sites contributing to the excesses ,- .:from digestive and respiratory cancer are greater than ex- shown in Table 5 for nonwhite males, Increased relatie -7: 100 Mortality Among Workers in Petroleum Refining and Petrochemical Plants/Thomas. Decoufle and Moufe-EfaSo ,".. .wquenciesof cancers of the stomach and lung are seen ngth-of-membership groups shown here. Results : & ,MR analysis are not too different from those of -e pMR analysis; however, the estimated risk for lung Jncer is reduced. In order to determine whether the increased risks of x i f i c cancers among OCAW members employed in eeoleum refineries and petrochemical plants were con- .dto employees of particular plants, mortalitv patterns ere examined in five plants From which a t least 100 eaths had been reported. Plant A is a large petroleum refinery with 421 deaths mong whites and 54 deaths among nonwhites. Some ?ion members employed a t this plant are guards; ,wever. most are production workers The observed dmber of stomach cancer deaths is significantly greater -an expected among whites who died 20 or more years -:er joining the Union (Table 9). The excess o i deaths '3m cancers of the brain and central nervous system is nit4 to workers with at least 10 years of membership. mong nonwhites there were one death from stomach 3ncer and no deaths from cancer of the brain. Plant E is a petroleum refinery and petrochemical iant that also produces sulfuric acid. Certificates were xovered for 413 deaths reported by the Union local dur?g the study period. Observed numbers of deaths for mcers of the stomach, skin, kidney. and leukemia are 'eater than expected for whites with 10 or more years of ?embership (Table 10). Significant excesses of stomach nd I t cancer are present among nonwhites. Expected u m t Jf deaths were not calculated for nonwhites by -ngth of membership because of the small numbers. :rain cancer was not excessive in either racial group. Significantly increased relative frequencies of brain (5 bsewed; 1.5 expected) and of lung cancer deaths (22 bserved; 11.7 expected) were seen among 198 whites $ h o worked at Plant C, which functions as a refinery and r d u c t s terminal. Although numbers were verv small, we -mined mortality patterns by length of membership nd found that the brain cancers clustered among males < h owere members for less than 10 years (all of the deceen& in this group with brain tumors were members for [ne years) while the lung cancer excess was confined to .orken with at least 10 years of membership. No ex3 ~ ofs specific cancers were seen among members Thoworked in the other two plants that were examined eparately )imssion T h e present study provides an example of the epidemih i ? i cvalue of data collected routinely tor administrative xrposes. As pointed out earlier, the primary use of this YPe d study is as a hypothesis-generating report. PMRs for some causes or death may appear higher +an they should because of lack of intormation on nembers who retired or lett the Union for other reasons. Conrlusions can be drawn with respect to exposure to ' N i f gents, however, categories of industrial p r e ':esses and individual plants associated with higher than risks can be identified. In spite of limitations im- by the "healthy worker effect" and the lack of ex30sure information, the finding that risks of stomach Journal Of Occupational Medicine/Vol. 22. No. 2lFebruary 1980 3 L (V0I 101 Lung (162. 163) '0 = Observed deaths t E = Expected deaths based on proportionate mortality for U S nonwhite males $E = Expected deaths based on proportlonate cancer mortality for U S. nonwhite males E = Expatea deaths based on proportionate mortality !or Texas nonwhite males 1Statistically significant at the 0.05 level -Tabb 9. Mortalii for Selected Cancer Sites in White Male OCAW Members Employed at WnI A by Length of Union Membership. Sie (8th Revision. ICD) I f 0' < 10 Ywn (N = 107) E t PMR Length of Membership I 10-19 Ywrt (N = 177) 0 E PMR I 0 2Oc run (N = 137) E Stomach (151) Brain & central nervous system (191. 192) 3 1.4 - 1 2.2 - 5 1.3 0 0.8 - 5 1.5 3.32$ 4 1.1 '0 = Observed deaths t E = Expected deaths based on proportionate mortality for U.S white males SStatlstically significant at the 0 05 level -Not calculated PMR 3.76$ - * - 4 Sir (8th Revision, ICD) Stwnach (151) Lung (162. 163) Skin (172. 173) Kidney (189) Leukemia (204-207) L White Mrhr Length of Membership < 10 Ywrs 1 0 t Ywo (N = 111) (N = 191) 0' E t PM R 0 E PMR 1 1.7 - 6 2.0 2.97$ 5 1 4.5 0.4 1-11 13 3 12.1 0.8 1-.08 -1 0.5 3 1.o - -3 0.9 - 3 15 Nonwhite males (N = 111) 0 E PMR 7 2.1 3 . w 13 4.4 2 . w -0 0 0.1 0.3 - -0 0.4 cancer, cancer of the brain, leukemia, and multiple mortality from brain tumors has previouslv been myeloma increase with longer Union membership among workers in the refinery and petrochemical category strengthens the possibility of an association with occupational factors. Furthermore, essentially the same relatiom reported. Although no positive gradient in risk with length of Union membership was detected for cancers of the P&R creas, liver, lung, and skin, further evaluation of these ships were seen in a proportionate cancer mortality analysis. These findings are consistent with reports of an excess of stomach cancer deaths among Canadian petroleum refinery workers,' and of an elevated risk of leukemia sites should be made. PMR analysis using Texas males I a comparison group does not alter the conclusions; h fore, these findings cannot be explained by high TW rates. The lack of a gradient raises the possibility that factors other than occupational exposure may be operating; associated with exposure t o benzene and perhaps other however, occupational factors cannot be completely petrochemicals.'] I s Studies of vinyl chloride workers have shown an increased relative frequency of brain tumors;'5 '' however, no association between employ- ruled out because some cancers may have shorter latent periods, depending upon the nature of the exposure. In addition, this study has highlighted three specific ment in refineries and petrochemical plants and excess plants in which Union members appear to have increased ' 102 Mortality Among Workers in Petroleum Refining and Petrochemical Plants/Thornas. Decoufle. and Moure-irag I risks of certain cancers - of brain cancer at Plants A and C, of stomach cancer at Plants A and B, and of lung cancer at Plants B and C. Further evaluations of mortality an workers at these plants will be made. w p i t e limitations in data and in methodology, results of ehe present study suggest that workers in this industry are at increased risk of certain cancers. Further studies are necessary to support and clarify these findings. The authors wnh to thank DR. I F Fraurnent. I r . R. Hoover. A. Blair, I . wJrath and MI 5 Krekel for support and guidance 5. Mayfield. T Prante. M. Z m . and D. Warn for technical assistdnce. and M. Clasrer for manuscript JrrgUltKn. References 1. Blot Wl. Brinton LA. Fraumeni IF, and Stone 81: Cancer mortalitv n US counties with petroleum industries. Science 198:jl-53,1977 2 . Hendricks NV. Linden N J ,Berq CM, et al- Cancer of the scrotum 'n wax pressmen A.M.A. Arch Ind Health 19524-529.1959 3. Wcil CS. Smvth HF. and Nale TW Quest for a suspected in- j u s u l a l carcinogen A.M.A. Arch Ind Hyg 5535-547,1952. 4. &id VC. Effects of atmospheric contamination on cancer mor- : a h in petroleum refinem exployees. / Occup Med 9.415-420.1967 5 TabenhawiCooper Associates: A mortality studv of petroleum dim workers. Project OH-1 prepared for the American Petroleum !nstitute. Washington, D C.. 1974 6. Wade L: Observations on skin cancer among refinery workers Arch Environ Health 6730-735.1%3 7 Hanis NM. Stavraky KM. and Fowler JL: Cancer mortality in oil refinery workers. / Occup Med 21:167-174.1979. 8.Considine DM (Ed.): Chemical and Process Technology Encycle pedia. New York: McCraw-Hill, 1974. 9 International Classification of Diseases (Eighth Revision). Geneva: WHO, 1%7 10.Monson RR: Analysis of relative survival and proportional mortality Compur Biomed Res 7325-332,1974. 11. Mantel N and Haenszel W . Statistical aspeca of the analysis of data from retrospective studies of disease. / Natl Cancer lnst 22719-748.1959. 12.McMichael AI. Havnes SC. and Tryroler HA: Obse*ations on the evaluation of occupational mortality data. / Occup Med 17:128-131.1975. 13 Vigliani EC and Saita C.Benzene and leukemia. N Ens// Med 2711872-8761.964 14.Infante PF. Wagoner JK, Rinsky RA, and Young RI: Leukemia in benzene workers Lancet 276-78.1977. 15.McMichael A I . Spirtas R. Kupper LL. and Gamble I F Solvent ex- posure and leukemia among rubber workers: An epidemiologic study. I OCCUP Med 17.234-239,1975. 16.Monson RR and Peters JM:Proportional mortality among vinvl chloride workers Lancet 2397-398.1974 17 Tabershaw IR and Caffey WR. Mortality study of workers in the manufacture of vinyl chloride and its polymers. / Occup Med 16.509-5181.974. 18.Waxweiler RI, Stringer W, Wagoner JK, et at: Neoplastic risk among workers exposed to vinyl chloride. Ann NYAcadSci 271:40-48, 1976. Meaning of Work For many persons, work is a vital part of the process of coping with life stress. This has two apsects: first, without work, the potential for boredom and meaninglessness is immeasurably increased. Work is often the primary means by which a person feels useful in life and through which significance and personal identity are established. The young adult must find a useful and meaningful place in society, and only in this way does a man or woman develop a firm sense of identity. Second, work is often a form of coping and a refuge. Consider, for example, how people in grief regard work. Many report that in the process of mourning and reorient- ing their lives after a major loss, work is not a burden, but the best refuge against cow stant high levels of distress and depression. One speaks here of "burying one's self in work." Thus, work may provide a psychological haven against problems that otherwise would be insurmountable, or against loneliness and depression. For such people nor to work is to be deprived of the only viable means of coping with non-work-related stress. - From Work Stress bv Alan A McLean AddisowWeslev Publishing Companv. Readink Massachusetts. 1979 Journal Of Occupational Medicine/Vol. 22. No. 2/February 1980 - 103