Document 4mgJgnxbxzM1bXyeNYDJMGXp

REE\!4LUATION OF BENZENE EXPOiURE FOR THE PLlOFlLM (RUBBERWORKER) COHORT (1936-1976) D. J.Paustenbach ChemRisk, A Division of McLarenlHart, Alameda, California P. S. Price ChemRisk, A Division of McLarenlHart, Portland, Maine W. Ollison, C. Blank American Petroleum Institute, Washington, D.C. I. D. jernigan, R. D. Bass ChemRisk, A Division of McLarenlHart, Springfield, Missouri H. D. Peterson Bryan, Cave, McPheeters & McRoberts, Washington, D.C. The Pliofilm cohort is the most intensely studred group ol workers chronically ex. posed to benzene. lnforrnafion on fhis cohort has been the basis for regu/ations and/or guidelines lor occupafion~al nd enrironmenfalexposure f o benzene. Rinrky et a / . (7986. 19En and Crump and Allen (1WdJeveloped different approaches lor reconsfrucfing the exposure hisfory of each member of the group. The predicfed levels o f exposure, combined with rhe data on the incidence of disease. have been used 10 esfirnafe benzene's carcinogenic potency. In f h i s paper. recent information horn worker inferviews and hisforrcal records lrorn the National Archives and elsewhere werc used to evaluate the accuracy o i prior exposure estimates and to develop better ones fsr $he cohort. The following factors were accounted for: (1) uprake of benzene due lo short-term, high-level exposure to vapors, Q) uptake due to background con. centrations in f h e manufacfuring building, (3) upfake due to contact with the skin, (4 morbidify and mortality data on workers in (he Pliofilrn process, IS) fhe installation o f indusfriat hygiene engineering confrolr, (61 exrraordinarily long work weeks during rhe 7940s. dafa indicating (hat airborne concentrafions of benzene w e r e underesti. mared due f o imccurare moniforing devices and the Iack of adequate held calibraffon --. . Requests for reprints should be Sen1 to Dennis J. PauslcnbJch, ChcniKisk. A Division of hlcLarenlHan, 1135 Atlantic Avenue, Alameda. CA 9401. 177 journal o f Toxicology and Environmental tieallh, 36:177-211, 1992 Copyright 0 1992 by Hemirphcrc Publishing Corporation \ / / I D.I.PAUSTENBACH E l AL. mated due to inaccurafe moniforing devices and the lack ofadequare field calibration of these devices, and IS) likely effectiveness of respirators a c d gloves. Our estimafes suggest rhaf G u m p and Allen 11984) overestimated the exposure of workers in some job c b s s i f i c a f i o n s and underesfimafed ofhers, and f l i a f R i n s k y et al. f7987, 39861 a / . most certainly underestimated rhe exposure o f nearly all worLers. Airborne concentra. tions of benzene af fhe Sf. Mavs lacilify during f h e years of i f s operation were found (on average) Io be abour half those I;/ fhe nvo Akron facilities. Our analysis indicares fhaf short-ferm, high-level exposure fo benzene vapors and dermal exposure signifi- c a n f l y increased ( b y about 25-50% the foul absorbed dose of benzene for some workers. O n e of rhc key findings was fhaf, unlthe prior analyses, fhe fhree facilif,es probably had significantly different airborne concentraflons of benzene, especially during the 79405 and 7950s. P INTRODUCTION The health hazards of human exposure to benzene are well docu- mented (Agency for Toxic Substances and Disease Registry, 1989; Gold- stein, 1977, 1983; IARC, 1982). In humans, exposure to high levels of ben- zene i s associated with a variety of disorders of the hematopoietic system, including leukemia and aplastic anemia (Coldstein, 1977; IARC, 1982). Evidence of the leukemogenic properties of benzene in humans comes both from collections of case reports (Aksoy, 1980; Greenburg, 1926; Hamilton, 1929, 1931; Vigliani, 1976) and from epidemiology studies (Infante et al., 1977; Ott et al., 1978; Rinsky et at., 1981; Wong, 1983; Yin et al. 1987). Unlike most chemicals, federal regulatory agencies in the United States have used epidemiology studies, rather than animal bioas- says, to derive cancer potency factors for benzene (OSHA, 1987; U.S. EPA, 1985). The Pliofitm workers are among the most intensely studied group in the history of occupational epidemiology (Cody et al., 1991; Crump and Allen, 1984; Infante and White, IS83; Infante et al., 1377; Kipen et al., 1988, 1989a, 1989b; Lamm et al., 1989; Rinsky et al., 1931, 1986, 1907; Ta- bershaw and Lamm, 1 9 m . The cohort consists of workers a t three difier- ent facilities in two cities in Ohio (one facility in St. M a r y and two in Akron). This group of rubbenvorkers i s regarded as superior to other epidemiological cohorts for understanding the hazards of benzene be- cause there are r e a s o n w o o d exposure data, there i s a lack of sipnifi- cant exposure to chemicals. employment and work history data are available on individual e , there is evidence of a dose-rvDmse relationqkp in the c o r n dehioiogv methods can bp ~~QLEctoL t b o u p (Brett et al., 1 A, 1985; Vo).tek and Thorslund, 1991). The studies conducted by Crump and Allen (1984) and Rinsky et al. (1986, 1987) estimated the workplace concentrations of airborne b,enzene. These estimates have been used as the basis for predicting the cancer --- BENZENE EXPOSURE OF PLlOFlLM COHORT 179 risk to humans (Albert, 1979; Austin et al., 1988; IARC, 1982; OSHA, 1985; Rinsky et al. 1987; White et al., 1982). Risk assessments based o n this cohort have played a major role in setting the current occupational expo- sure limits for benzene as well as numerous air, water, and soil standards (American Conference of Governmental Industrial Hygienists, 1991; OSHA, 1987; U.S. EPA, 1985). cpIn light of t h e irn ortance of this cohort, w e conducted a search of historical records an conducted interviews with workers to identify ad- ditional information that might reduce the uncertainty associated with prior estimates of t h e workplace concentrations made by Kinsky et al. (1986, 1987) and Crump and Allen (1984). This paper critically evaluates the two previous efforts and presents information not considered in ei- ther estimate. A significant source of the information contained in the paper comes from i n t e w i e w with. t h r w St&n I < errlp14) tees (Allman, 1991; Hanning, -1990, 1991a, 199lb; Osborne, 1991). These individuals had from 25 to 39 years of workplace andlor manaqement exagLiPnrp at the St. Marys tacil- ity from 19-12-1976. O n e of the individuals had been employed in the Akron mI-af 1934-1942.While many employees of St. Marys were interviewed by earlier investigators, it is uncertain that they were ques- tioned concerning their experience during the early years of production (1936-1950). These are almost certainly the years of greatest exposure. We present a n approach that quantitatively accounts for the follow- ing factors: (1)short-term, high-level airborne concentrations of benzene; (2) dermal exposure and uptake; ( 3 ) longer work weeks during the 19.10s and 1950s; (4) t h e quality of the available air sampling data; (5) improved local exhaust ventilation, and (6) the less-thancomplete effectiveness of -respirators. New estimates of worker exposure were developed and com- pared with previous published values. Our analysis indicntcs that c x p - Sure a t t h e facilities was significantly g r e x r in the 194- rly 1950s than previously reported by Rinskv et al. 1(- 'k.e those predicted bv Lrump and Allen (1984L BACKGROUND INFORMATION The Pliofilm Manufacturing Process Pliofilm was a strong, thin, flexible material similar to plastic wraps now commonly used in home food storage. T h e chemical process of Pliofilm manufacture was described in detail by Rinsky et al. (1981) based o n a survey by t h e National Institute of Occupational Safety and Health (NIOSH). In 1976, several NIOSH investigators conducted a walk-through survey of t h e Pliofilm operations at the St. M a y facility (Young e t a!., 1977). From this survey, NIOSH developed a schematic diagram (flow chart) of the Pliofilm manufacturing process. We have revised that dia- 180 D.1. PAUSTENBACH ALl. gram (Fig. 1) based on information gathered in recent worker interviews (Allman, 1991; Hanning, 1990, 1991a, 1991b; Osborne, 1991). Figure 2 presents our best reconstruction of the floor plan of the St. Marys facility (194G-1976). The primary differences between the I ~ V OdiJgrams are the addition of several operations, including (1)the scale tank between the blend tank and the reactor to weigh the materials so that the proper amounts would be used; (2) the scrubber unit after the reactor and neutralizer, which was used to remove the gases released during the reaction; (3) centrifuge between the quencher and the storage tank (it separated unwanted particles and broke the waterhenzene emulsion); and (4) the finishing and shipping operations. The di3grari.i hclps explain how the airborne concentrations for one job category (or areas in the plant) may have been related to those for another. The production process can be summarized as foilows. Natural rubber was disso!ved in benzene and converted to rubber hydrochloride by the addition of hydrochloric acid. After a ripening period, the unreacted hydrogen chloride in the mixture Li'as neutralized with sodium carbonate and the resulting slurry was filtered and stored. The solid materials filtered from the solution were transferred as a cake to a "quencher" unit where residual benzene was separated and recovered by steam distillation. The rubber hydrochloride solution, often referred to as cement, was turned into film on a castin unit. During this process, the benzene content of the film was reducec f from 90% to 10%. Evaporated benzene was swept out of the casting unit into a wet scrubber that cooled the vapoi and removed residual hydrogen chloride before discharging the benzene-laden air to charcoal absorbing unit. The benzene was recovered and reused. The resulting Pliofilm w a s further dried in a drying unit and stored on rolls before moving to finishing operations. In the finishing operations, the film was rerolled, cut, and stretched before packing and shipping. Benzene levels in the dried Pliofilm i s believed to -have been negligible. Those processes known to have produced especially high airborne concentrations of benzene were (1) mixing the rubberibenzene solution, (2) neutralizing t h e solution, (3) filtering the solution, (4) quenching the solution, (5) adjusting the thickness of the Pliofilm in the casting unit, and (6) gauging tanks of the cement in the storage room. The highest d.zgree of exposure generally is believed to have occurred when the filter press (a plate and frame filter) was cleaned and during operation of the quencher (Hanning, 1990a). These particular unit operations, which are not unique to Pliofilm manufacturing, h a v e been known for decades to be associated with high exposure. Previous descriptions of the Pliofilrn Operations have divided the op- eration into two sections called the "wetside" and the "dryside." The definition of these terms has varied (United Rubber LVorkers, 19TT). Workers at St. M a r y s defined the "wetside" as t h e operations that oc- BENZENE EXPOSURE OF PLIOFILht COHORT 181 1: "l.,, *,.c,:t .O .,"'..*:'r FIGURE 1 . Pliofilm process, SI.M a r y ' s . Ohio curred before the casting of the film. Dryside operations included rubber hydrochloride storage, casting, finishing, and shipping. Workers on the wetside were considered separate from the dryside workers and reported to different operations managers (Allrnan, 1991; Hanning, 1990; Osborne, 1991). In this paper, we adopted the St. Marys definition: that is, workers M'ho were involved in precasting operations were considered wetside, and all others were dryside workers. The Pliofilm Manufacturing Facilities Pliofilm was produced at three facilities (see Fig. 3) in two locations, 300 miles apart (Rinsky et al., 1981).The first facility (Akron I) w a s located in Akron, Ohio, and it began commercial production in 1936. It was originally an experimental pilot plant, which was expanded into commercial production. It was located-on the second and third floors of a multistory factory building. The facility shared t h e second floor with a balloon fabric manufacturing operation that also used large amounts of benzene. The second Pliofilm manufacturing facility w a s opened in the 1930s at St. hlarys, a small town 300 miles southwest of Akron. The St. Marys facility w a s built to meet production needs for Pliofilm as demand increased at the end of the Depression. The St. Marys facility i v a s located in a oneand two-story building adjacent to a larger manufacturing facility. In 19.18, the third facility (Akron 11) opened in Akron a few miles away from 1a2 0Y n I. i D. I . PAUSTENBACH ET A t . FIGURE 2 . Pliofilrn plant, St. Mays, Ohio Akron I. This facility replaced the initial one (Akron I), which shut down in the same year. -- Description of Previous Exposure Estimates To date, two attempts have been made to reconstruct the jobspecific exposure history of the Pliofilm cohort (Crump and Allen, 1984; Rinsky et al., 1986). Both studies are based on the available industrial hygiene rnonitoring data presented in Rinsky et al., (1981); however, the two studies reached significantly different conclusions regarding the airborne concentrations of benzene to which the workers were exposed. Both approaches used the extensive job history records available for Pliofilm workers. Those records specify the job title, as well as the starting and finishing dates of the job assignments for each worker. There are 30 jobs titles listed for the St. Marys workers and 64 for the workers at the two Akron facilities. In many instances, several job titles refer to the same worker activity, or to activities with similar potential for benzene exposures. To simplify the exposure assessment, both Crump and Allen (1984) and Rinsky et al. (1986) collapsed the job titles into approximately i 183 1 8 4 D. I. PAUSTENEACH E l AL. 10 exposure categories, which they believed represented similar types of activities and exposure. Airborne concentrations of benzene were estimated for each of the job categories for each o i the years during which the facilities were in operation. Industrial hygiene monitoring data were used whenever available. Finally, the individual workers' exposure histo- ries were reconstructed from the individual workers' records (by job) and the benzene levels for each exposure category. Estimates of the likely concentration of benzenr kzpors for each ex- posure category differed significantly between Crurnp and Allen (1984) and Rinsky et 21. (1986),with the former ususlly having higher estimates. The differences in benzene concentrations result from differences in the way each researcher estimated the workplace concentrations for the years prior to 1963 when there were limited monitoring data. While more than 2000 measurements of the airborne concentration of benzene are available for the three sites, the majority of the values comes from the St. Marys facility, and these values were collected after 1969 (see Fig. 3). Only a handful of air samples was collected at St. M a y s prior to 1956. The two Akron facilities have less data than St. M a r y s . The results of 400 shortterm (length of stain tube) samples are available for the A k r o n I t facility (in operation from 1939-1965), but it i s not known in what years they were collected (Mclnerney, 1977). The results of only three samples are available for Akron I . Faced with this absence of data for 1936-1956, both Rinsky et al. (1906) and Crump and Allen (198.1) were forced to make numerous assumptions when developing their estimates of airborne benzene. In this paper, w e address the differences in the exposure factors they selected and evaluate each of them in light of information not considered when their analyses were conducted. NELV INFORMATION ON EXPOSURE We identified new information invol\ping seven factors t h a t co6m Sig- nificantly change prior exposure estimates for the Pliofilm workers: ('1) I I ( accurac; of air;amplin devices, (2) length of the work week, (3) rubber ;,/ is .us durrnq V h r l h l~~~~~ ''I 5, (5) / / i' /' I aermal uptake, (6) effectiveness 6f overexposure. of rcspiratr:lrs, and (7) m--e~dical widencc I naccurat e Mon i torins Dcviccs Before 1963, airborne concentrations of benzene were determined b m using one of two analytical methods: detector tube kit? and combustible gas indicaprs (Rinsky et al., 1981, 19S6).Verbal accounts by industrial hygienists who worked in the industry confirm that these were used in the Pliofilm building (Mclnerney, 1977; Rinsky et al., 1981). Detector tube kits consisted of a package of two unmixcd reagents, empty BENZENE EXPOSURE OF PLIOFILh\ COHORT 185 glass tubes, and a squeeze-bulb aspirator. just prior to use, the reagents were mixed and the tubes were filled and capped to form the detectors. An aspirator was used to obtain an air sample of a fixed volume of 33 ml nominal capacity (Rinsky et al., 1981). Benzene concentrations were pro- portional to the length of stain produced when air \vas drawn through the tube (Rinsky et al., 7981; Silverman, 1956). Several authors have assessed the reliability of measurements using these detector tubes (Kusnetz et al., 1960; Silverman, 19561, and e a c h has listed their deficiencies. Specifically, the accuracy of detector tubes was evaluated by Hay (1964).H e conducted a laboratory experiment where h e sampled known concentrations (from 25 to 250 ppm) of benzene vapor. found that the detector kit gave readings that were, on average, only of the actual value at a benzene concentration at 25 ppm. That is, the detector kit displayed a reading of 15 p p m when the actual benzene &concent tion in air was 25 ppm (Hay, 196-4). This underreporting (by about ivas fairly consistent for benzene concentrations ranging from 25 to 100 ppm. When the tubes were used to measure concentra- tions above 100 ppm, the underreporting increased. For example, at con- centrations of 250 ppm, the tubes reported concentrations of o n l y 109 YrnH -.m64). In c o n t r s t , a subsequent field evaluation by Hay s owed t h a t tube readings overestimated the airborne concentration of benzene thought to be present in a plant t h a t refined coke-oven light-oil products. Hay attributed these disparate results to interace from toluene, xylene, thiophene, and phenols present in the manufacturing process (Hay, 1990). Fortunately, the solvent used in Pliofilm production during the 1930s was about 95% benzene, with the major contaminant being toluene (JViIson, 1942). Based upon Hay's work, we conclude that it is appropriate to adjust reported levels of benzene obtained by this method by a factor of 1.5. The second analytical technique for measuring workplace concentra- tions of benzene vapor (prior to 1963) was the combustible gas indicator (CGI) (Rinsky et al., 1981; Wilson, 1942).The CGI relied on the catalytic combustion of gas or vapor over a heated platinum filament and the subsequent imbalance o f an electrical resistance bridge. Pagnatto et al. (1961)evaluated this device in a study of benzene exposure in the rubber coating industry. The authors noted that, when r ar t l y~ ope~rated, th,e CCl reported onlv about half the actual benzene concentration b h e n Ct 4 v h e r e airborne concentrations of benzene ranged from 5 to 125 ppm. In addition to evidence provided by Pagnatto et al., there are two other reasons why the benzene levels were likely underestimated by the CGI. First, the wire used in the CGI's electrical bridge loses sensitivity with age (Silverman, 1956). Second, field calibration of the CGI used in the Pliofilm operations ivas often rudimentary during the 1940s-for ex- 186 D.I.PAUSTINEACH ET AL. ample, employees frequently used a Zippo lighter to determine whether the instrument was operating (Hanning, 1990; hlcinerney, 1977). Based on this information, benzene concentrations measured at t h e St. M a y s facility before 1963, and at the A k r o n I and II facttlities, were likely to have been underreoorted by a t least a factor of 1.5. Where such -`kedmeasurements are cited in this paper, wCIrl8lcale value and the corrected figure in square brackets [ I. The values in brackets were used in our estimates of worker exposure. Ex t ended Work Peri ods There i s considerable evidence that the work week at t+e Pliofilrn facilities was often much longer than the modern .M-h schedule. Due to the n a t u r e of the process, Pliofilm typically was manufactured on a continuous basis. During the initial years of Pliofilrn production, Depression- era unemployment caused industry to use four 6-h daily shifts as a means of providing jobs for a greater number of people (Hanning, 1990). Employees typically worked 5- or 6 d work weeks. However, in the 193Os, with the beginning of U.S. involvement in World Lt`ar II, production in the rubber industry greatly increased and labor was in short supply (Department of Labor, 1942). Employees began working 8 h shifts, and extended work periods were not uncommon (Hillman, 1942; Rothstein, 1932; Zimmer, 1931). Extension of the individual work week to 50 or more h and the payment of attendance bonuses were being considered in 1943 in an attempt to meet wartime needs (Flanick, 1943). Similar work patterns were used in St. Marys where employees worked 6 to 7 d a week (Hanning, 1990; Osborne, 1991). In the late 193Os, employees continued to work more t h a n 40 h per week. The typical Pliofilm production schedule a t St. Marys .iovolved operating for 19 d with a 2d shutdown for maintenance. People were expected to work one 8 - h shift, 7 d a week, including weekends (Osborne, 1991).This information is consistent with the data in the personriel records. For example, a workman's compensation report for a person diagnosed with leukopenia indicated t h a t he worked 49 h per week for an entire year (Industrial Commission of Ohio, 1944). The change in the average work week during the 1930s and early 1950s can be seen in the records of hours worked per Lveek recorded in the accident and injury reports, see Table 1. The information shows t h a t three out of four employees (75%) for whom data are available worked more than 40 hlwk during the period between 1940 and 1911, with a n average work week of 46 h. The proportion of employees working overtime remained approxi- Rmately constant throu h 1951, but the average work Lveek increased to 49 h. In 1952-1956, t e percentage of workers ivith long schedules dropped to 4076, and the average work week was decreased to 44 h . It is acknowledged that those who filed an accident or injury report may not BENZENE EXPOSURE OF PLlOFlLM COHORT 157 TABLE 1. Hours Worked per Week by Various Members of Pliofilm Cohori for 1940-1956 feriod 19-a-1931 1942-1936 1947-1951 1952-1956 40 h/wk 1 3 4 6 AS hlwk 3 6 5 3 56 hlwk . 0 3 5 1 Average (hlwkl 46 48 49 44 ~ Number of workers with >40 hlwk ~~ 314 (75%) 91:2 05%) 70114 V2%1 4'10 (40%) Note. Based on dafa obtained from applicable Coodyear "Foreman's accident and injury re. pons" (dated 1937-1956). be representative of the entire work force. Nevertheless, these data, coupled with the testimony of workers and knowledge of the pressures on similar industries during the war years, give u s confidence that long work weeks were common. The alternative to using t h e injury report data would be to use the worker interview data and sei the work week at 51 h per week (see discussion on uncertainty analysis). \W'll Shutdown of St. Marys fPliofilm w a s in reat demand by the military for protecting e q u i p ment from dust an moisture during shipment to the war zones. Hoivever, with the beginning of hostilities in the Far East, supplies of rubber were decreased and the ability to produce Plioiilm was affected. For example, from 1942-1945, the St. Marys faci1i:y received little new rubber from Indonesia and relied on the recycling of old Pliofilm (Hanning, 1990).Although recycling efforts were intense, Pliofilm production at the St. Marys facility was reduced to 4 7 % of capacity. Evidence for a decrease in production can also be seen in a report of the history of the St. hlarys facility published in an employee newsletter Wingfoot Clan, 1983).In contrast with St. Marys, the Akron I plant continued to produce Pliofilm at or near capacity during 1942-1945. The reduced production at St. Marys could have been satisfied by using one, rather than four, casters. Even this caster needed to operate only o n e shift, rather than three per day. The unused capacity in St. hlarys was devoted to the production of coated, waterproof tent fabric and papedmetal foil laminate (Hanning, 1990; LYingfoot Clan, 1983). Neither of these products used benzene as a solvent (Hanning, 1990).Worker's job histories d o not reflect these changes. The records indicate only that workers were operating the neutralizers or casters. They do not specify what products the equipment was manufacturing (Hanning, 1991b). The lack of availability of rubber for St. Marys from 1942-1945 is expected to have lessened worker exposure to benzene for se-deral reasons, First, the time workers spent in the production of Pliofilm (and thus ' -. 108 D.I . PAUSTENBACH ET AL. were exposed to benzene) w a s reduced from threc to one shifts per day. Second, the reliance on recycled Pliofilm eliminated the need for certain steps in i t s manufacture. For example, the mixers and the reactors were not operated since the chlorination step was not performed (scrap Pliofilm was added directly to the neutralizers). Third, the number of times the filter presses were cleaned and changed w a s reduced greatly since sodium carbonate was not necessary to neutralize excess hydrogen chloride. Based on this information, it was concluded that the background concentration of airborne benzene in the facility was relatively low and that many of the workers in the exposure categories, which would have had high levels of exposure, were exposed to relatively low concentrations (e.g., about 5-20 ppm). Medical Evidence of Elevated Exposure in the Early Years The major difference behveen the Crump and Allen (7984) and Rinsky et al. (1986) analyses was that the former concluded that worker exposure to benzene was significantly higher in the early 1940s than in the 1960s. Some scientists at OSHA who e1,aluated the Crump and Allen (1984) anal- ysis believed that human exposure to airborne concentrations in excess of 150 ppm for 8 h!d for weeks or months was very unlikely because these would have produced acute illness (Crump and Allen, 1983). At that time, Crump and Allen and OSHA were not aware of information that indicated that incidents of acute toxicity had been observed in the Pliofilm cohort. We have since found published accounts that demon- strate the existence o f significant hematopoietic effects, including fatali- ties, in the Pliofilm workers in tht. 1930s. In addition, there i s ample evidence in the industrial hyaiene literature that some fraction of work- ers can be exposed to 100-%2ijO pprn for l o n v periods f6 h/d for several weeks develooin clear sienz o f A r l p c 5 f ' P 4,19j1:-'GTpen- uro, 1926). e m a 1932 Department of Labor Conference on Health H$ards in t h i Rubber 1ndustr):'Dr. Conn, hledical Director of the Goodyear Hospital at Akron, described the results of a study of Cood- year workers who routinely had exposure to high levels of benzene: EIn tests of 1,104 employees exposed to more than 11501 p a r t s p e r million of benzol over any considerable period of time, 7.5 p e r c e n u 3 ) s a d a qli- F 1 ' . . TLVOand two-tenths percent (75) (all men who had- E r million showed matked eblood chan es. Nine serious disturbances of the 00 - orming organs that had a sufficiently severe aplastic anemia to require hospitalization. Of these nine, three died. Those surviving have returned to work. They are tested frequently and a r e apparently in good condition. Uivo fatalities came from ttle spreader room: one from Pliof i Im .) BENZENE EXPOSURE OF PLlOFllM COHORT 189 Dr. Wilson, a physician with Goodyear, reported on the same group of workers, stating that while their exposures occasionally could reach 500- 1000 [750-1500] ppm (such as when vapors were released while opening a drum of benzol or from the heating of benzoltontaining media), rou- tine concentrations varied between 50 (75) and 500 (750) ppni, y i t h "aver- age concentrations being about 100 (150) pprn" (Wilson, 1942). It i s clear trom t h e reference to a fatility that occurred in the Pliofilm department and from later references to Pliofilm operations in the tran- script (Department of Labor, 1942) that some members of the Pliofilm work force were included among the 1104 workers exposed to benzene concentrations of 100 (150) ppm or greater. However, it is also clear that the survey included workers manufacturing other rubber products such as balloon fabric. Therefore, we cannot conclude that all cases occurred among the Pliofilm workers, nor imply that the reported incidence rates of hematopoetic effects apply to the Akron I Pliofilm workers. A second incident where exposure to benzene during the 1940s i s known to have produced hematopoietic effects in Pliofilm workers oc- curred at St. hlarys. Injury reports submitted to the Ohio Industrial Com- mission indicated that two neutralizer operators at the Pliofilm facility suffered from serious blood abnormalities in July 1946 (American Petro leum Institute, 1986). Both were hospitalized, and one eventually died. The information in these reports suggests that the industrial hygiene program facilities in the Pliofilm in the 1930s \vas not sufficient to pre- vent acutely toxic exposures to benzene. In evaluating this information, it should be remembered that benzene workplace concentrations of .IOO-lSO \ w r P n m - n o n in nitmpro115 5- durin.g these eca de i K ok , 19.15). The clearest evidence for high benzene exposure, hoitcever, i s pro- vided in the rprQrds of the St. L l a r y s ivorkers that have been evaluated b y m e t ai. (1989a). These workers were monitored monthly (red and white blood cell counts) in order to manage the em- ployee risk due to exposure to benzene. Figure 4 represents the average white blocd cell WBC) counts by year for the St. M a y Pliofilm workers. I t sh0M.s that for the years 1930-1949, the workers' average ll%C counts -were su.p.pressed. However, durinR the 1950s the counts remained stable. h i n 51 to benzene concentrations in the rant'? n i ? j - [ ) 111. Apparently, airCorne levels of benzene decreased tKroughoutz9:Os such that bv /* the 1950s the levels were low enough to stppression of \YBCs no longir occurred. The findings of Kipen et al. (1989a) have been questioned by Hornung et al. (1989). They suggested that the temporal trend in blood counts could also be explained by factors other than exposure to ben- I. 190 D. I . PAUSTENBACH T AL. UI I . 10' .1 C I -i -t +++ + +++I +T . 8 I 4 """"""""'~ 1940 194 5 1950 1955 *.are I I, I,,..,,, 1960 1965 1970 FIGURE 4. The average white blood cell count for Pliofilrn workers at SI. M a r y s during the years 1930-1970(Kipen et a!., 198951. zene, since preemployment measurements of blood counts were lower in t h e early 19Ws than in later years. They also suggested that the test method used at that time could well have been flawed. Kipen et al. (1989b) published responses t o these claims wherein t h e y showed that the trend in preemployment data was insufficient to discount the dramatic change in the blood counts. Cody et al. (1991) have also shoxn that the preemployment data from the 19-10s could be used to demonstrate a progressive suppression of blood counts during the first 6 m o of employment in t h e Pliofilm process. Evidence for Improved Engineering Controls Estimates of exposure by Crump and Allen (1984) and Rinsky et al. (1981) have been based on the assumption that the Pliofilm manufacturing process remained relatively unchanged from the time of its develop ment in 1936 until its discontinuation in 1976. Our research indicates that although the chemical process probably remained relatively constant, t h e three facilities had different manufacturing equipment and industrial hygiene controls. Further, it is well known in the industrial hygiene community that employee exposures are not strictly a function of the operation but are also a function of work practices, room size, ventilation rate, building layout, and local exhaust ventilation (Mclnerney, 197i9. These BENZENE EXPOSURE OF PllOFllM COHORT 191 characteristics are known to have differed from facility to facility and to have changed over time. The following describes the differences in the facilities and the known engineering changes in the manufacturing process at the two facilities for 1930-1970. In 1936, the process for manufacturing Pliofilm was developed in Akron (Akrcn I). The location of the first production facility was a portion of the third and fourth stories of a multistoried building in which a number of rubber products were manufactured. Pliofilm production and balloon fabric manufacturing were on the same floor (Department of Labor, 1912; Hanning, 1990). Only general division ventilation was provided. A second Pliofilm production iacility was built in the early 1940s in a two-story building in St. Marys, Ohio. Unlike Akron I, this facility was designed explicitly for the purpose of manufacturing Pliofilm. While the processes and equipment were similar to Akron I , the building layout was signiiicantiy different. First, the ventilation for the building was provided by large ventilation fans located in the basement of the building that supplied filtered air (for dust removal) to each room of the building (Hanning, 1991b). Second, the finishing operations, which were located in a separate building in the Akron facility, were in a room between the \vetside operations and the casters (Beebe, 1976). In the early 194Os, a shortage of toluene caused many rubber operations throughout the industry to shift to benzene as a solvent. These changes resulted in widespread cases of leukopenia among Akron rubber workers (including Pliofilm cohort members). In response, benzene was replaced with other solvents in all of Goodyear's Akron manufacturing departments except Pliofilm and balloon manufacturing. In the balloon fabric area, the Akron management instituted engineering controls including improved local ventilation, and in both departments they maintained a program of monthly and, later, quarterly blood monitoring (Department of Labor, 1932). In November 1946, following the cases of benzene toxicity observed at St. M a r y s , management apparently embarked on a major program to reduce airborne concentrations of benzene (Fluker, 19-16).The engineer- ing controls included local exhaust ventilation around manivays at certain tanks and placement of the wetside filter presses within ventilated enclosures. Local exhaust ventilation was provided at the hatches of the mixer, reactor, and neutralizer units (Hanning, 1990). Pliofilm production required that rubber be added manually into the mixer. Scrap Pliofilm, plasticizers, additives, and sodium carbonate were added to the neutralizer vessels. While adding these materials, workers were exposed to relatively high concentrations of benzene vapors displaced from the tanks (Hanning, 1990). The plateand-frame filter presses, which removed the sodium carbonate and sodium chloride from the rubber hydrochloride solution after the neutralization step, required regular cleaning. This cleaning involved the scraping of caked salts off the filter cloths. The / J J192 D. I. PAUSTEYBACH ET AL. cleaning process placed the neutralizer operator directly over material wsaturated with warm (9O0-12O*F)rubber hydrochloride-benzene solution ( 9 s / be ene) for approximately 30 min e a c h time the press was c eane T h e local exhaust ventilation kras probably effectibe at pulling a substantial amount of the benzene vapors away from the workers, but typical airborne concentrations for such operations in other industries can be as high as 500-1000 p p m (see Table 9). n-'\ia single air sample was collected at the filters before the engineering controls were installed, and the concentration was above 250 (375) pprn. Samples collected several years later (after engineering controisj showed levels of 19-50 (27-75) ppm (Rinsky et ai., 1981).The state of Cihio concluded t h a t -the filter-press ventilation system was sufficient to maintain benzene concentrations below 100 ppm and vpically below 35 (52) pprn (Fluker, 19466). After 1946, additional ventilation improvements were made in the casting units at St. hlarys in response to a series of flash fires within them B(Allman, 1991; Hannin , 1991).The effect of the ventilation on the work- place concentration o benzene is not known. I t is acknowledged, however, that testimony by members of the labor union indicates that local exhaust ventilation equipment was not always kept in repair JUnited Rubber Workers, 1977). The general (dilution) ventilation system that was designed into the St. Marys process was the first engineering control measure installed at a Pliofrlm facility ( H a n n i n g , 1990). it is not clear if such controls would have been installed at Akron I, since the Akron I I facility was buil: 2 yr later. A limited industrial hygiene survey of the Akron I facility in 1948 did not allude to the use of ventilation, and the concentration of benzene reported in the area where the rubber hydrochloride solution was stored was 500 (750) m (Fluker, 1948). Engineering controls similar to those insta4 e a q . s are believed to have been installed a t X l r o n I I (Ha n ni ng, 1991a). The Akron I t facility opened in 1948 and incorporated a significant change in the design of the casting u n i t . Casters at the Akion I and St. M a y s factilities passed heated air over the Pliofilm to remove the benzene. The benzeneladen air then passed throtlgh a n activated charcoal recovery system. Benzene concentrations in the casters were kept below t h e lower explosive limit by the large volumes of heated air that passed through them. Large exhaust fans kept the casters cinder negative pressure. At Akron 11, the heating and removal of benzene were achieved by radiant heaters. Benzene concentrztions were maintained above the u p per explosive limit, and an inert gas was introduced into the casters. YBenzene was recovered b condensation on a refrigerated coil. This sys- tem was apparently more ikely to leak benzene t h a n the use of activated carbon, and the odor of benzene was reported to be detectable around the Akron I I casters (Hanning, 1990). In contrast, the odor of benzene I BENZENE EXPOSURE OF PLlOFlLM COHORT 193 was not reported to be detectable around the St. Marys casters (Hanning, 1990). The threshold of smell for benzene is somewhere between 25 and 100 ppm (ASTM, 1978; Dr3ger, 1980; Cerarde, 19631, depending o n smoking status and other personal characteristics. This value is consistent with measurements of 35 (50) ppm benzene reportedly found near t h e Akron I1 casters (Rinsky et al., 1981). Dermal Exposure In addition to inhalation exposure, dermal contact by workers a p pears to have been common. The contribution of dermal exposure t o benzene uptake was acknowledged but not quantitatively accounted for in the previous ejcposure estimates of Rinsky et al. (1987) and Crump and Allen (1984). Based o n worker interviews and other records, we have been able to identify several points in t h e Pliofilrn process that required employees to have significant dermal contact with benzene. The available information indicates that Pliofilrn workers rarely, if ever, used gloves (Allman, 1991). Even if they had, based o n what is known about gloves used in such operations for the years 1940-1970, only a modest level of protection would have been afiorded the workers since nearly all gloves manufactured at that time were at least partially permeable to benzene. Recent work has shown that virtually all gloves available before 1985 had a very limited period during which they prc- vided protection (Barnardinelli and Bender, 1990; Nelson et al., 1981; Weeks and McLeod, 1982). The dermal uptake of benzene due to dermal contact has been stud- ied by several researchers. Hanke et al. (1961) showed that liquid ben- -zene passed through the arms of human volunteers at a rate of 0.4 . Research in animals has reached similar conclusions (Franz, eMaibach and Anjo, 1981). The work by Hanke et al. (1961) is be- lieved to be the best o n e to use to estir,iate the doses received from dermal exposure since the rate was determined in workers' forearms over a period of 1.5-2 h. Pliofilm workers Lvere dermally exposed t o benzene on the hands and forearms for periods of time up to 1.5 h. Based on Hanke's et al. work, o n e can estimate the dermal uptake of benzene by considering t h e number of contacts and the duration of contact with benzene cement solutions: -Dermal uptake P x N x A x t x RA\' (11 where dermal uptake is the absorbed dose (mg/kg/d), P the molar concentration oi henzene i n the henzene/cernent solution, N the number of skin contacts per d, A the surface area of contacted skin (cm'), t the contact time (h/d), R the dermal absorption rate (mglcrn'lh), and \Y the body weight (kg). I 1-33 0.I . PALJSTENBACH E l AL. Example Calculation Workers who were responsible for the neutral- izers are believed to have had their hands and forearms in contact with liquid benzene for about 30 min per shift while cleaning the ivetside filter press. The likely dermal uptake received during this task may have been approximately: -p -N -t -A -R -\Y - -Dermal uptake 0.95 1 timeld 0.5 h 1980 cm2(surface area of hands and forearms) 0.4 mglcm'lh 70 kg (0.95)(1)(l983)(0.5)(0.4)/70 5.3 mg/kg/d An absorbed dose of 5.3 rnglkgld of benzene is appreciable compared to t h e uptake d u e to inhalation. Accepting an overall average of 50% inhala- tion absorption over 8 h at concentrations of 10-100 ppm (Nomiyama and Nomiyama, 19741, an inhalation rate of 1.2 m'lh, and a conversion factor of 3.2 mglm' per ppm of benzene, t h e airborne concentration of benzene necessary t o produce an equivalent dose over an 8-h shift would be about 25 ppm. Table 2 presents estimates of uptake for those exposure categories believed to have had significant dermal uptake. The following equation estimates the airborne concentration of benzene that for an 8-h workday produces a dose equivalent to that of thc dermal exposure: -Airborne concentration (CAC) (by x CJ!(A` x R x B ) (2) where I.\' is the body weight (70 kg), U the dermal uptake (rng/kg/d), K the conversion factor (3.2 mglm' equals 1 ppni benzene), R the respiratory rate (9.6 m'l8-h d), and B the inhJlation bioavailability (50%).-ln--our assessment, t h e EAC was added to the estimate of concentration inhalation t o calculate the total average daily equivalent airborne concentration. The development of dermal exposure estimates involves consider- able uncertainty. To evaluate the uncertainty in the estimates of the EAC values, a Monte Carlo model of the equation was developed. Variation in dermal exposures arises from the differing degree of absorption of benz e n e through the individual's skin, the extent of contac:, and the frequency of contact. Hanke et at. (1961) reported the mean intake and confidence limits for the rate of absorption based on the variation found in his test subjects. The variation in the extent of skin area in contact with t h e cement is based o n our conclusion that the tasks would have involved contact of the hands and some portion of the forearm. The variation in the area of contact ranges from a minimum of the hands to a maximum of the hands and forc.arrils. Ttlc frequency of performing the 195 196 D.1. PAUSTEKBACH ET AL. activities that resulted in dermal contact (i.e., cleaning the filters and operating the quencher) varied from shift to shift. Based on worker inter- views and information on the process, we concluded that the operations could have been performed from 1 to 1.5 times a shift. Table 3 presents the actual assumptions used in the Monte Carlo model. The results of the model are presented in Table 2. These results show that the impact of the variations on the estimates of EACs in Table 2 i s relatively small (a factor of 2) and the values used in this paper fall within the 95th and 5th percentiles of the estimated distribution. Based on the result of the model, we conclude that while there i s some unceriainty in the estimates of dermal exposure, the contribution from this source i s significant and should be included in the analysis. Changes in Some .Exposure Categories Interviews of the workers and a reanalysis of the records indicated that certain jobs should be reassigned to different exposure categories than those assigned by Rinsky et a!. (1986). In addition, these data allowed us to better relate the available air monitoring information to t h e exposure categories. For example, company monitoring data taken around the casters were reported by several subgroups using terms such as "behveen units" and "platform" (Rinsky et al., '1981). It is now clear that the workers in the exposure category called "knifeman" spent the majority of their time where the "platform" air samples were taken, and the workers in the category of "caster operator" were located where the "between units" samples were collected. Our analysis indicates that exposure of t h e finishing workers to benzene has been underestimated in previous studies. These workers were involved in finishing operations that included rerolling, cutting, trimming, tensilizing (stretching the film to increase i t s strength-and reduce i t s thickness), packing, and shipping. Management apparently did not monitor for benzene in finishing areas. Crump and Allen (1984), as well as Rinsky et al. (1986), believed that finishing workers had minimal expo- TABLE 3. Sources o i Uncenainty Considered in the hronte Carlo Model of Dermal Exposcre Range Area o f skin exposed 840-1980 c rn' Rate of absorption 0.35-0.41 rng!cmld Frequency of exposure 1-1.5 timeslshift Dis: r ibut ion Uniform Tr ia nguI a f (most likely value 0.30) 1) nIl or rn 8a 5i s Exposure w a s assumed to involve borh hands and part of t h e forearm Hanke e t at. (1961) 1 IVorker interviews on frequency of press cleaning and quencher operation BENZENE EXPOSURE OF PLIOFILM COHORT 197 sure to benzene. Crurnp and Allen (1984) assigned finishing workers time-weighted average WP,)concentrations of 1-5 p p m . Rinsky et a!. (1986) assigned exposures of 0 pprn for the workers. h`orkers who were only employed in the finishing area were excluded from the Kinsky COhort (Rinsky et al., 1981). Based upon the following evidence, we concluded that the workers in the finishing departments of St. M a y s and Akron II weie exposed to low but significant airborne concentrations of benzene. First, the floor plan of St. Marys (Fig. 2) indicates that the finishing operations were located in rooms between the wetside and dryside operations. Thus the workers in such areas were almost certainly exposed to fugitive ernissions of benzene from both operations, Second, the University of North Carolina (UNCI report (1974) indicated t h a t odors from operation in one part of the plant were readily detectable in other portions of the facility, thus suggesting t h a t air exchanges betLveen rooms were significant. For example, in a discussion of the mixing operation, UNC reported that "This operation and the subsequent batch drop mill generates considerable smoke and a rather unpleasant odor ivhich permeates the plant." When discussing the Prime Wrap process, a product produced in the Pliofilm department in St. Marys during the 1970s, Uh`C noted that "the distinctive odor of tetrahydrofuran (the solvent used in the Prime Wrap DPuilding:'rocess) was quite evident at various levels throughout the [Pliofilm ] Third, UNC took four time-weighted average samples from the finishing operation areas and reported values of 0, 12, 20, and 30 pprn benzene. Interestingly, these levels are significantly higher than benzene concentrations measured by UNC in many of the wetside and dryside work areas (UNC, 1974). In contrast with the finishing workers at St. Marys, Akron I workers were not exposed to fugitive emissions of benzene, since these opera- tions were located in a separate building (Beebe, 1976). Therefore, the Rinsky et al. (1986) assumption of low exposure for this group is probably correct. The location of the finishing operations at Akron I I i s unknown. However, Akron 11, like St. h4arys, was designed for Plioiilm production. It is therefore plausible that Akron II used t h e integrated design of St. hlar).s and located the finishing operations in a rooin (or area) adjacent to the casting units. The conclusion that fugitive emissions could result in significant background levels also led to the reconsideration of certain job titles that require that a worker move about in the facility or spend time in the facility in areas where monitoring data are unavaila!)lc. 1-or such workers we developed three new exposure categories: general wetside, general dryside, and facility average. Estimates of exposures for these categories are based on averages of the background monitoring for the operations in the wetside, dryside, and facility as a whole. The finistiing operations --_ 198 0.I. PAUSTENBACH fT AL. for Akron I I and St. Marys facilities were assumed to be in the general dryside exposure category. One key finding from our investigation involves the estimated con- centration of benzene in the dryside area at St. hlarys for the years prior t o 1946. Crump and Allen (1984) concluded that workers in these areas were exposed to airborne concentrations of benzene in excess of 250 pprn. Their estimates were based on their understanding t h a t the casting units were not enclosed and ventilated prior to 1936. Apparently, they confused the ventilating of the casting units with the improved ventila- tion of the wetside filter presses that occurred t h a t year. Although it is true t h a t additional local exhaust ventilation was installed at the St. Marys facility in 1946, the evidence is that the casting unit was always enclosed and vented to a solvent recovery unit (Hanning, 1990). It is highly unlikely that benzene concentrations around the casting units regularly approached or were in excess of 250 ppm. ESTIMATING WORKER EXPOSURES In this section, the key differencesin the Crump and Allen (1984) and Rinsky et al. (1986) assumptions a r e contrasted and the merit of these differences is discussed in light o f the new data. Table 4 summarizes the differences in the hvo approaches and the analyses presented in this article. Estimating Exposure for 1930-1970 The major difference amoj3g the estimates derived by the three r e search groups involves certain assumptions for the 1940s and 1950s. Crurnp and Allen (1983) assumed that the workpface concentrations for each job category were related directly to the increased knowledge about benzene's toxicity. Specifically, they believed t h a t exposures stead- ily decreased as the ACGlH threshold limit values (TLV) decreased from the 1940s to the 1970s. D u r i n z t h e vea r s t h a t Pliofilm wi15-Dreduced, there were six C ~ I W P ~:i thc h T I \ -5). The TLVs ranged from a high of 2C)Opm (befprp 1937'! I of 'IC, m m (1970 and after). Therefore, Crurnp and Allen (1984) divided the exposure es6mates into six periods of time corresponding to when the various TLVs were in place, plus a seventh period for the years before the first occupational exposure guidelines were drafted (1946). If there were no reported mea- surements for a n exposure category (E,) during a specific time period 01, they estimated the airborne conccntration for t h a t period based on rnea- +surements taken in a later period (i -t- 1) times the ratio of the period's TLVs (5, S,+,). That is, 199 200 0.I. FAUSTENEACH ET AL TABLE 5 . Changes in t h e Threshold Limit Value ULV) for Benzene for the Years o l Pllofilrn Production s 1946 1547 1948 19-49-1957 1958-1%3 1964-1969 21970 100 100 5.0 M ' A 35 TWA / ' 25 T W A 25 Ceiling / 10 nvA / For example, the exposure category "quencher" had no reported measurement for the year 1930. The TLV at that time was 50 ppm. The average measurement for the period of 1949-1957 was 39 ppm and the TLV was 35 pprn. Therefore, Crump and Allen (1984) estimated that the exposures for the year 1348 were - -,,,E, 39 ppm (50 pprn/35 ppm) 56 ppm Using a similar analysis, the exposures in 1947 were estimated to be 111 PPm. In contrast, Rinsky et at. (1986)assumed that no change in workplace concentrations occurred unless specific information was available to show otherwise. They used only time-weighted average airborne samples for the year in which the measurement was taken to determine i f any change had occurred. For the years in which no data were available, exposures were interpolated from previous and subsequent values. For the years before data were available, they assumed that the value of t h e earliest year was applied to all years before it. For example: reported measurements for the mixer exposure category were available for the years 1949 (45 ppm) and 1956 (10 pprn). Based on these values, Rinsky el al. (1986) interpolated exposures for the years 1950-1955 and set all years before 1949 equal to 45 ppm. Thus the estimates of the exposure for the mixer operator were: 1956 1955 1954 1953 1952 10 ppm 15 ppni 20 ppm 25 pprn 30 ppm 1951 35 ppm 1950 40 p p m 1949 45 ppm <I943 45 ppm 7As we have discussed, there i s medical, en ineering, and general histori- cal evidence to support Crump and AI en's (1983) assumption that worker exposure in the 1943s was higher than during the 1950s and 1960s. Th. question, howevt.r, i s lvhcthc!r i t i s Iikcly that workplace concentrations of benzene dt.crcased at the rate that corresponds to changes in the TLV. There a:(: several reasons why this approach i s reasonable. First, data from industrial sunq's of the St. Marys Pliofilm departmeni (the sc-called "112 Surveys") clearly indicate that workplace concentrations were consistent with the 1'369 change in !he benzene TLV (Rinsky et al., 1981). That is, samples collected in the areas where workers spent much of their time alreraged 13 pprn for the years 1963 to 1967: a value about half of the prevaiiing TLV of 25 ppm. The surveys taken from 1969-1973, in the same areas, averaged 5 ppm (again about half the p r e vailing TLV). Second, it i s clear that some degree of compliance with the TLL" ivas important to the managers of the Pliofilm operations. For example, FIuker (1948) and Sefarian (1956),both with the State of Ohio Department of Industrial Safety, make reference to 1LVs in their reports and letters on Pliofilm operations. Third, the "112 Surveys" were begun in 1963, the year that the ACCIH TLV for benzene changed from a t i m e Bweighted avera e value of 25 ppm to a "ceiling value" of 25 ppm. Ceiling valves w e r e de ined as concentrations never to be exceeded during the work day. We believe that the decision to conduct a large survey of work- place exposures at the time the TLV changed to a ceiling value is not coincidental, but rather was a practice frequently undertaken by major firms concerned about meeting occupational standards. Finally, evidence that the Crump and Allen (1984) approach i s reasonable is that these estimates are consistent with the limited measurements taken in the 1940s and 1950s (see Table 6). One shortcoming of using the TLV ratio technique t@estimate exposures during the 1930s i s that it can produce unlikely estimates for certain years. For example, in 1943 the TLV for benzene dropped by a factor of 2 , so our estimates of concentration io: the dryside workers in St. h 4 a y dropped in a similar factor. This drop in workplace concentration may or may not have occurred that rapidly. interviews of workers from this period do not indicate that any significant changes in benzene con- trols were installed on dryside equipment (Aliman, 1981; Hanning, 1990; Crisrner, 1991). A more reasonable view i s that sometime during the 1950s, the airborne concentration benzene levels in the Pliofilm operations decreased to meet the 1948 TLV. As with any major undertaking, several years would be needed to install engineeiing controls to reduce u'or kp1ace concent rations. Evidence That Monitoring Data Do Not Reflect Peak Exposure The second major question about how to retrospectively estimate employee exposure i s whether the grab samples (which make u p the bulk of the available data) adequately reflect the short-term peak expo- 202 D.I . PAUSTEhiBACH ET AL. TABLE 6. hleasured concentrations of Benzene Vapor and Concentrations Estima!ed Using the TLV Ratio Technique Year 1941-1942 1% 1947 1949 1950 1956 Location in plant' All areas Wetside filter presses Storage Storage Mixers Storage Mixer Reactor Neutralizer Storage Reponed concentrationb (PPm) >loo >35 19 1G 45 35 0 0-35 0-35 0-15 Concentration adjusted for analytical bias' (PPm) I> 150) [>SI] 1291 1151 167 1=1 [Ol 10-521 [O-521 10-261 Background concentration (ppm) bawd o n the 112 Surveys data multiplied by t h e TLV ratiod 60-114 58 25 25 60 25 15 1 20 25 Reference Depanment of Labor (1942) Fluker (1946) Rinsky et al. (1981) Rinsky e! al. (1981) Rinrky et al. (1981) Rinrky et al. (1981) Sefarian (19563 Sefarian (1956) Sefarian (1956) Sefarian (1956) 'Location in SI. hlarys facility where value was measured. 'Results of industrial hygiene samples of benzene a! SI. Marys. 'Measured airborne concentration of benzene multiplied by 1.5 to correct for inaccurate instrumentation. dBackground values determined by taking the average of the 112 Surveys data from t h e years 1963-1967 and multiplying by the ratio of the TLVs. These valves have not been adjusted for peak exposure, dermal exposure, or extended work weeks. sures of the workers. Crump and Allen (1984) did not account for peak exposures but believed that some effort should be made to address them, especially for the casting operators. In contrast, Rinsky et al. (1986) believed that grab samples were representative of the peak concentrations and claimed that these measurements w e r e collected selectively in those areas where benzene exposure was the highest. Therefore, they believe that these samples overestimated the background concentrations within the building. While it is acknowledged that the St. Marys industrial hygiene program probably focused on those activities that were likely to produce the highest concentration of benzene, it i s not clear that the samples were taken at times when benzene exposure would have been at i t s highest. Colorimetric tubes used in the "112 Surveys" and Akron II surveys only reflect the concentration of vapor over a time period of 30 s to 2 min. For the results of surveys using these tubes to accurately reflect the 8-h TWA Eworkplace concentrations, a careful sanipling rogram based on time- weighted worker activity patterns would have een necessary. There is n o evidence of such a program. It i s unlikely that samples were collected during the period of greatest exposure, since during the 1977 OSHA BEKZENE EXPOSURE OF PLlOFlLhl COHORT 203 hearing o n the revised benzene standard, Pliofllm workers testified t h a t they were instructed not to enter areas where respirators were required when wearing sampling devices (United Rubber Workers, 1977). Additional support for concluding that the grab sample data fails to adequately reflect peak exposures can be found by comparing "112 Surveys," estimates with the 4- and 8 h JWA measurements (Rinsky et al., 1981). For example, the "112 Surveys" data for the reactor operators in 1973-1974 indicate that the TWA airborne benzene concentration was less t h a n 1 ppm, whereas the 8 h TMA data indicate that concentrations were 9 pprn in 1976. Similarly, the "112 Surveys" estimate for neutralizer operators in 1973-1974 indicate a concentration of approximately 3 ppm, while the a h TWA data for the s a m e years suggest 25 ppm. Finally, samples collected at St. Marys in the 1970s at t h e employees work station found concentrations of 98 and >200 ppm at t h e dryside plate and frame filters during cleaning, and 100 and 200 pprn during the gauging and inspection of tanks (Rinsky et al., 1981). The levels found in t h e "112 Surveys" for areas near tanks assumed to be closed averaged 1.7-15 ppm, and at t h e dryside presses the values averaged about 70 ppm. Based o n these data, we concluded that the Akron I I and "112 Surveys" were taken in t h e areas of high potential benzene exposure but not necessarily during t h e time when benzene exposures were at their highest (such as during sampling and loading tanks). Therefore, we believe there is sufficient evidence that g r a b sample results are representstive of the general background levels oi benzene in the building and usually do not reflect short-term peak Concentrations. Evidence That Akron I, Akron II, and St. hZarys Had Different Workplace Concentrations Both Rinsky et al. (1981) and Crump and Allen (1983) concluded t h a t the similarities of the processes at the St. h4arys and the two Akron facilities allowed air sampling data from a n y site to be used to estimate expo. sure at all three facilities Vable 4). For example, Rinsky et al. (1987) noted that "Processes and joh assignments were essentially identical at both [Pliofilm] locations, so benzene exposure levels measured at [St. Marys] were assumed to be naturally occurring simulations of exposure levels in corresponding areas at [Akron], when actual exposure measurements did not exist." Based on the available information, this assumption no longer appears plausible. First, as discussed above, there were significant differences behveen the three facilities as far as room size, layout, and design of machinery. Second, a review of the two largest sets of air sampling data, the "112 Surveys" at St. Marys and the survey taken at Akron 11, shoivs that the data from Akron II are appreciably higher t h a n those of St. Marys. Finally, there is biological evidence for suspecting that the employees . ... 204 D.1. PAUSTENBACH AIl . at the three sites had different levels of exposure. Lamm et al. (1389) reported that the leukemias observed in workers at the Akron facilities were of a significantly different nature than those at St. Marys. h4ore importantly, Rinsky et al. (19111) reported that the standard mortality ratio (SMR) for the cohort at Akron I and II was about double that of St. Maty (746 vs. 345). This difference is stili apparent in the 1990 update by NlOSH (Paxton, 1992). Justificationfor Averaging Years The Crump and Allen (1984) approach for estimating workplace concentrations was dictated by their belief that "benzene exposure would be approximately the same unless standards (e.g., TLVs) were changed." Therefore, these authors averaged the air sampling data collected over the period during which the TLV w a s constant. The approach used by Rinsky et al. (1986) to estimate exposure was to assume that measure ments of benzene primarily applied to t h e year in which they Lx'cre taken. We chose to adopt Crump and Allen's (19843 TLV ratio approach in our analyses. Description of Our ApproJch Based on new information collected and the above discussion, we have developed revised estimates of the airborne concentrations of ben- zene for the various exposure categories at the three facilities. In dcveloping the estimates, the assumptions were made: I 1. The Akron I , Akron I I , and St. Mary's facilities were different. We estimated the airborne concentrations at Akron I I based on the 400 air samples actually collected at the Akron I I facility. The "112 Surve)gs" data collected at St. M a r y s were used to estimate eiiiosures at that site. However, because of the virtual absence of monitoring data at the Akron I facility, the "112 Surveys" data were also the basis for the Akron I estimates. 2. The ratio of the workplace concentration of benzene to the prevailing TLV was assumed to be constant. Worker exposures at St. Marys were estimated by calculating the mean of the "112 Surveys" results for the years 1963-1968 and dividing by the TLVs for those years. For example, Rthe estimate of the back round concentration for the workers in the neutralizer category for t e 1950s was estimated as follows. The mean value of the "112 Surveys" for these workers was 15 pprn for the years 1963-1968. This value was multiplied by 1.4 [the TLV during the 1950s (35 pprn) divided by the TLV for 1963 (25 ppm)]. Thus our backcalculated estimate of the benzene vapor concentration around the neutralizer for the 1950s was 21 ppm. The ratio of 1.4, with the exceptions discussed below, w a s assumed to be consistent for the years BENZENE EXPOSURE OF PLlOFlLM COHORT 205 between 1949 and 7957 for all jobs at St. M a r y s . This approach was not used to estimate airborne concentrations at Akron II because the exact yearb) in which the measurements were made are unknown. 3. The Akron and St. M a r y s "112 Surveys" provide plant-wide averages rather than peak exposures. As discussed previously, the averages of the "112 Surveys" and Akron I I suwey were assumed to be generally representative of background concentrations, that is, the prevailing building concentrations at the work station rather than peak exposures. The approach used in this paper was to reconstruct the dura- tion and frequency of peak exposure activities based on interviews of workers, and to add these to the background concentration. Tables 7 and 8 present a summary of our estimates. The benzene concentrations that occurred during high exposure activities were estimated using one of two approaches: (a) the upper end of the distribution of concentrations reported in the "112" and Akron I I surveys, and ( b ) monitoring information on peak exposures from similar industrial processes (see Table 9). The survey data were used to estimate the peak concentrations at St. hlarys for the years after 1936, and at the Akron It facility when local ventilation was in place 10 control peak concentrations. 'The Akron I facility was assumed to have similar exposures to St. Marys for 19.10-1938. Estimates from similar processes were used for Akron I and St. M a r y s for the years before 1946. A factor to account for the use of gas canister masks (respirators)was also included in the analyses. The masks when worn were assumed to be able to reduce the inhaled concentration by a factor of 4 (see discussion that follows). Table 10 and Table 11present the estimated duration and concentrations for each of the peak e x p e sures. 4 . Adjustment of early monitoring data for ana!ytlcal biases. Measure- ments of airborne concentrations of benzene taken before 1963 Lvere underreported by the instrumentation. The values were therefore adjusted-upward bv a factor of 1 5 . 5. Airborne concentrations of benzene at St. hlarys in the 1940s were affected by the shortage of natural rubber and the installation of sig- nificant local exhaust ventilation. During the years 1942-1945, Pliofilrn produc;ion at St. M a r y s ivas dependent predominantly on the use of recycled Pliofilm, and production was limited to 4-77; of capacity. This implies that, on average, workers were exposed much less frequently and that the background concentrations were fairly low. However, due to uncertainty as to whether the total number of workers in the Pliofilm department was less during the war years, we assumed that exposures occurred during onefourth of t h e work week. The use of recycled Pliofilm eliminated the need for operators in the mixing, mill, and reactor jobs (Hanning, 1990). We assumed that workers in these areas received only minimal exposure to benzene, 206 D. I. PAUSTENBACH E7 AL. TABLE 7. Estimated Airborne Concentrations of Benzene Used to Calculaie Uptake by Pllofilm Workers i n Akron I or II Exposure categoryldate Background Concentration (ppm) Peak 1 Peak 2 Peak 3 Peak 4 Mill 1936-1946 53 1947 27 leu] 19 1949-1%5 39 Mixer 1936-1946 53 1917 27 1948 19 1949-1965 39 Reactor 1936-1946 4 1947 2 1938 19-49-1965 24 Neutralizer 193&19& 53 1947 29 1% 20 1949-1965 30 Quencher 1936-1956 113 79-47 57 194a 40 1949-1%5 59 Knifernan 1936-1946 79 1947 40 1938 28 1949-1%5 59 Spreader 1936-1946 76 1947 33 1940 27 1949-1965 54 Still house 1936-1946 43 1947 20 1948 14 1949-1965 3.3 -` N A Not applicable KA' N A NA Nh NA NA N A N A N A NA NA N A N A N A NA NA 200 N A N A K .4 150 t4 A N A N A 105 rJ4 N 4 N A 75 K 4 N4 N A 2w 225 N A N A 10 N A N A N A 7 N A NA NA 75 NA K 4 NA 250 200 225 15 7 20 K 4 10 04 N 4 150 150 N A NA KA NA NA 300 200 200 40 300 120 140 20 300 04 98 14 -_300 150 150 38 - 120 250 250 N A 120 250 250 N A 120 250 250 NA 120 250 250 N A 250 750 K A 250 750 h A 250 750 h A 250 62 N A NA NA NA NA NA NA NA NA NA N A NA NA NA N A K A N A NA NA NA NA BENZENE EXPOSURE OF PLlOFllM COHORT 207 TABLE 8. Estimated Airborne Concentrations of Benzene Used to Estimate UptaLe by the Pliofllnl Workers in SI. M a r y s Exposure categoryldate Background Mill 19Jo-1931 53 7942-1945 0 1% 53 1947 27 1948-1956 19 1957-1968 13 1%9-7972 6 7973-1976 9 Mixer 1949-19 41 53 19-42-1945 0 19% 53 1947 27 1948-1956 19 1957-iw 13 1959 6 1970-1972 6 1973-1976 9 Reactor 1940-19: 1 53 1942-1945 0 1946 53 1947 27 1948-1956 19 1957-1965 1 1%9- 1972 2 1973-1476 7 Neulralizer 1940-1941 58 19-42-1945 58 1% 58 194: 29 1948-1956 20 1957-1968 15 1969 5 1970-197? 5 1973- 1976 10 Quencher 194-1946 113 1947 57 1958-1956 40 195?-19G 2a 1969 1970-1972 14 14 1973-1976 9 -'NA NOI applicable Concentration (ppm) Peak 1 Peak 2 Peak 3 NA' N A N A NA NA NA NA KA NA NA NA N4 N A NA N A NA NA KA NA NA NA NA NA NA 200 N A K A 0 N A K.4 200 N A N A 150 N A N A 105 NA N 4 75 K A N A 75 N A N A 0 N A h'A N A NA N A 200 225 N A 0 0 NA 20 N A N4 10 NA NA 7 NA N4 5 NA N4 5 N4 h4 h A h'A N 4 250 200 225 250 200 0 250 2Dc) 225 15 120 N 4 10 84 h'A 7 60 N .A 7 Go N A 3 25 K A NA N A NA 300 200 203 300 120 140 300 e-4 99 300 60 70 300 60 70 300 25 200 NA NA NA Peak 4 NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA N.A N.4 NA NA NA NA NA NA NA NA NA NA NA NA 40 20 14 10 5 5 NA 208 D. I. PAUSTENBACH E l A l . TABLE 8. Estimated Airborne Concentrations of Benzene Used to Estimate Uptake by Pliofilm Workers in St. Maryi (Continued Exposure categonfldare Knifeman 1ero-1946 le17 1948-1956 1957-1468 1%9-1972 1973-1976 Spreader 1940-1936 1947 1940-1956 1957-1W 1%9-1972 1973-1976 S t i l l house 1w-1942 1933-1945 1946 1947 1948-1956 1957-1968 1969-19R 1973-1976 Background 79 40 2a 20 5 18 76 3a 27 19 2 9 40 10 40 20 14 70 5 9 Concentration (pprn) Peah 1 Feak 2 Peak 3 120 250 250 120 250 250 120 250 250 120 250 250 120 250 250 NA NA NA 250 250 N A 250 35 N A 250 25 N A 250 18 N A 250 10 N A N A N.4 N A NA N A N A NA N A NA N.4 N A h'A NA NA NA NA NA NA NA NA N4 NA NA NA NA KA NA Peak 4 NA NA NA NA NA NA NA ' NA NA NA NA NA NA NA NA NA NA NA NA NA Finally, the frequency of wetside filter press changing was also reduced. Exposure estimates for Akron I were not reduced-Gce p r e duction information indicates t h a t the Akron facility operated near capacity throughout World War II (Crismer, 1991). The engineering improvements installed at St. M a y s in 1946 in- cluded local e x h a u s t ventilation at the wetside operations of mixer, reactor, neutralizer, and filter press. Estimates of peak exposure for the years before such controls were installed were based on data from similar processes. Estimates ot'the concentrations in the storage room were also adjusted to reflect the single measurement taken before 1946. We h a v e made the assumption that A k r o n I also installed controls in 1946. 6 . As discussed above, certain jobs h a v e been assigned to different exposure categories t h a n in earlier studies. Table 12 presents the revised assignments. 7. Dermal exposure should be accounted for in the estimate of the work- ers' total absorbed dose. As discussed, the contribution of dermal exposure to worker uptake can be estimated from the frequency and BENZENE RPOSURE Of PLlOFllM COHORT 209 TABLE 9. Approximate Breathing Z o n e Concentrations of Benzene for Various Tasks Frequently Performed in the Chemical Manufacturing, Pharmaceutical. a n d Rubber Industries Breathing zone concentration (ppm) Task Without LEV' With L E V Reference Sample reactor contents with ladle through a manway Cleaning a filter press Operating a centrifuge Loading a reactor through the manwsay Pouring from one drum into another 80-120 (100 p p m ) b 100-1ooo (250 ppm) 150-500 ppm (300 ppm) 100-300 p p m (200 ppml 100-200 ppm (100 ppm) 10-30 (20 p p m ) 10-Bo p p m (40 p p m ) 25-50 p p m (35 ppml 20-60 ppm (40 ppml 10-40 p p m (20 ppm) Bond (1931) Paustenbach (19m Bond (19911 Paustenbach (1977) Schoch (1991) Paustenbach (19m Dorsey (1973) Noie. The temperature of benzene was assumed to be 2OOC. Higher temperatures would necessitate correcting the data. Data in this table were obtained from various industrial hygienists w h o have collected samples during these tasks or unit operations. 'LEV, local exhaust ventilation. %alues in parentheses represent t h e author's best estimate of the most likely conccntration in the Plioftlm operation. TABLE 10. Estimated Duration of Background and Peak Exposures for Vvorkers at Akron I and II ~~ Exposure clars/date Background (hid) Peak 1 (hld) Peak 2 (hldl Peah 3 (hldl PeJk 4 Wd, h4ill 1936-1 9 6 8 h4ixer 1936-1965 7.08 Reactor 1936-1946 194;-1965 4 6 Neutralizer 1936-1946 1947-1%5 -4.20 6.20 Quencher 1936-1965 3 Knifeman 1936-1965 7.62 Spreader 1936-1%5 7 -' N A Not applicable NA' NA NA NA 0.92 NA N.A K A 2 2 NA NA 2 NA NA NA 0.50 1.30 2 0.50 1.30 N A NA N.4 1.33 1 1.67 3 0.13 0.17 0.08 NA 0.06 0.92 NA NA I' 210 D.I. PAUSTENBACH ET AL. TABLE 1 1 . Estimated Duration of Background and Peak Exposures for Workers a t SI. Marys Exposure class/da:e Background (hldl Peak 1 Wd) Peah 2 (hid) Peak 3 (hid) Peak 4 Wd) h\ili 1949-1976 Mixer 0 194+19?2 7.03 1973-1976 Reactor N.A 1940-1936 4 1947-1972 1973-1976 NeurrJIizer 6 NA 1940-1946 4.2 1957-1972 6.2 1973-1976 N.A Quencher 1940-7972 3 1973-1976 NA Knifeman 1950-1972 7.62 1973-1976 h'A Spreader 1940-1972 7 1973-1976 NA -'NA Not applicable. NA' 0.92 NA 2 2 NA 0.50 0.50 NA 1.33 NA 0.13 NA 0.08 NA NA NA NA 2 NA NA 1.30 1.30 NA 1 biA 0.17 NA 0.92 NA NA NA NA NA NA NA 2 KA NA 1.67 NA 0.08 NA NA NA NA NA NA NA NA NA NA NA NA 3 NA NA NA NA NA duration of dermal contact. The &h equivalent airborne concentration of benzene was calculated and added to the estimates of background and peak concentrations to obtain the total dose (uptake). 8. Exposure estimates should be normalized to an 8-h day, S-dwork week. The average length of the Pliofilm work week was approximately 36 h from 1936 to 1930 and then between 40 and 56 h until the mid 1950s.The impact of this finding i s that the employee was absorbing a much greater dose in the mid 1'340sthan the airborne concentrations would indicate. In order to account for this factor, estimates of exposure were normalized to a 4 - h work week. 9. Exposure estimates for 1973-1976 should be estimated based on the availability of higherquality data. In these years, three surveys of workers were made using either 4-h or 8-h n V A samples. Unlike the grab samples, these samples would be expected to include peak exposures. Therefore, these results were used where possible to determine uptake via inhalation. Exposure Calculations The estimates of airborne concentrations were based on the following equation: ii 21 1 212 EEKZEKE EXPOSURE OF PLlOFlLM COHORT 213 Exposure - +{I({Il ( R X Cn] X C D, x C,} D, x CJ18 h] + C EIU} X I 40 h per week (4) where D, i s t h e duration of t h e ith peak exposure, D,the dilration of the background exposure, C, the vapor concentration for the ith peak activ- ity, C, the vapor concentration for the background activity, R t h e respira- tor effectiveness, U t h e fraction of time the respirator is used during peak exposure, EIU the equivalent inhalation uptake from the ith dermal exposure, and H the hours worked per week. This formula accounts for the workers' uptake d u e to background levels, short-term peak expo- sures, dermal exposure, and for workweeks that varied in length. An example calculation is presented below for the neutralizer at St. Marys in 1936. Based on interviews and historical records, workers in this categov are believed to have performed several tasks that involved short-term, high-level inhalation exposures and to have performed two tasks with significant potential for dermal contact. To estimate the equiv- alent daily airborne concentration of benzene vapor, the following ap- proach was used: -D, 1.33 h -C, 200 ppm D, = 0.5 h D, - 2 h -D, 4.16 h -R 0.75 -EIU, 24 ppm C2 = 250 ppm -C, 250 ppm -C, 58 ppm -u 0.33 -EIU, 8 pprn H - 48 Airborne concentration - {[I - (0.75 X 0.33)) x [1.33 x 200) + (0.5 x 250) t - (2 x 225) + (4.16 x 58)]/8 + (24 + 8)) x 48/40 170 pprn (rounded to two significant figures) The estimates should be viewed as the average daily airborne concentrations of benzene vapor that would have to be inhaled for 8 h/d, for a 4 0 h work week, to be equal to the amount of benzene absorbed by these workers under t h e conditions of exposure. Tables 12 and 13 present our estimates for the workers at the two locations. U N CERTAI N N ANALYSI S It is.not possible to develop confidence limits or specific quantitative estimates of uncertainty for the airborne concentrations developed in this paper. The available analytical data are too limited and the estimates are too dependent on inference and indirect evidence to use techniques 214 D. I. PAUSTENBACH ALl. such a s Monte.Carlo or error propagation. However, it is possible to test t h e impact of the major assumptions o n t h e estimates t o identify those assumptions and factors that drive t h e analysis. This section presents a n evaluation of the impact of slight and major changes in the key exposure factors and their effect o n the final estimates. The exposure factors examined include: T h e Crump-Allen TLV ratio approach Lengthened work weeks 0 Possible analytical biases Use of protective equipment Uncertainty in the estimates of dermal exposure T h e relative impact of selecting various exposure factors was evaluated in t h e following manner. First, t h e estimated concentrations of each of t h e major exposure categories (knifeman, spreader, mixer, reactor, neutralizer, quencher, still house, and facility average) were averaged t o yield a plant-wide background average. This plant-wide average was calculated each year. Second, these yearly values were then averaged for the years Pliofilm was manufactured at the facilities in the two locations, Akron and St. h4arys. The Akron I and II facilities were combined in t h e analysis since the work force at Akron I was transferred to the Akron I 1 facility. Thus, the workers remained the same. The resultant location averages provide a useful measure of h o w each factor affects the worker exposure at each of the locations. In addition to these location averages, a n additional set of averages were calculated for t h e years 1940-1949. The purpose of this second set was to illustrate t h e effect of the factors o n t h e years when the least amount of data were available and when expcsures were believed t o be the greatest. The results of the -a-nalyses are presented in Table 1 4 . RES ULTS The airborne concentrations predicted in this paper ranged from 2 few parts per million to several hundred parts per million, depending on the exposurs category, location, and year. The facility average exposures at t h e two Akron facilities were higher than at St. Marys and were higher in t h e 1940s than at later periods. Our concentrations differ from both t h e Rinsky et al. (1986) and C r u m p a n d Allen (1984) estimates. Table 14 presents a comparison of t h e averaRe concentrations estimated by t h e i h r e e investigators. Our estimates ar;, depending o n location in the'facilities, t h r e e to five times higher than those of Rinsky et al. (19861. HOWever, our estimates are quile similar to those of CrumD and Allen (1984) for t h e St. Marys facility'and about 25% higher for th6 two Akron facilities. Figure 5 is a set of graphs of the three estimates for three different BENZENE EXPOSURE OF PLlOFlUI COHORT 215 TABLE 14. Estimates of the Equivalent 8 hld Concenlration (ppm) of Benzene for the Ohio Pliofilm Workers,' Cornparison to Previously Published Estimates, and the Eflect o! Usmg Alternative Exposure Factors Our estimates Comparison IO published values Rinsky et al. (1986) Crump and Allen (1984) Relative contribution of background peak, and dermal components Bachground only Background and peak Background, peak, and dermal Sensitivity analysis TLV approach N O adjustment for TLVb Adjustment for TLV Lengih of work weeh 40h 51 h' Analyticald h o adjustment for possible bias hiaximum bias reported (2 0) Respirator use' Use of respirators during all tasks' PNo use of res irators Dermal contact 5th Percentile 95th Percentile Akron I, II Akron I, II St. Mays (1936-7966) (1940-1949) (1940-1976) 86 106 45 23 25 19 63 103 44 St. h\atys (7%&1949) 63 7f 25 L3 B7lo3 54 67 23 69 87 30 06 106 45 43 50 63 71 74 38 06 106 45 78 08 4 1 90 113 4 7 78 101 45 93 107 45 77 93 41 94 118 49 77 96 37 89 110 49 43 63 53 68 63 63 57 69 56 b6 'An average of the exposure categories, knifeman (platform), spreader (between units), mixer, reactor, neutralizer. quencher. still house, and facility average. qL\{ratio approach (see text) i s not used. Background levels reported in 1963-1967 w e r e applied to a l l years before 1463 (Akron I and S t . Mays facility only). CLongest reported work week (Osborn, 1991). dCorrection of pre.1463 analytical measurements for analkqical biases. 'Respirators are assumed to be worn whenever the worker i s exposed to airborne concentra. tions in excess of the prevailing TLV. Respirators assumed to be 75% effective when worn. 'Results of the Monte Carlo model for dermal uncertainty (see Table 2 and 3). exposure categories at the two different locations. In general, our esti- mates tend to follow Crump and Allen's estimates more closely than Rinsky et al. (19861, and both are about fivefold greater t h a n thosegf Rinsky et ai. (1986) for most of the job titles. However, we differ from Crump and Allen's estimates in two ways; first, we found that the average concentrations at the two Akron facilities were higher t h a n their estimates, and second, we concluded that wetside exposures, represented in Figure 5 by the neutralizer, were higher t h a n for the dryside workers, represented by the spreader operator. 216 D.1. PAUSTENBACH ET A L -Neutralizer St. Marys D D I IO0 E Q U I I 60 V a n I '0 D D n 40 E 0 u I I 10 V n I '0 - -\ -Spreader Operator St. Marys i0 210 - 200 --- n 't 0 . I . . 1 , , . . . . . I . . RLP W HO IU W HI *PO Year I))) FIGURE 5 . Comparison of the predicted exposure profiles for three dLfferent jobs using the meth o d r of Rinshy et al. (19861, Crurnp a n d Allen (1984). and oursclves BENZENE EXPOSURE OF PLIOFILhi COHORT 6 ?OO h 0 v 1 1bO P P m E 100 Q u :I 6o n 1 -Neutralizer Akron I 8 II -Mixer Akron I & ll D P m 40 E a Y I I ao V W n I '0 nao V84I I**D Year W48 MO -Spreader Operator Akron I & II ltu 8 100 h 0 7b0 U I -200 P P m 0 h\a U I 100 1 a l 0 n WP W41 UW hU W 98W H1 FIGURE 5 . Comparison of t h e predicted exposure profiles for three different jobs using the meth ods of Rinsky et at. (1986).Crurnp and Allen (19W)a,nd ourselvcs. Y 218 D. I . PAUSTEh'BACH Ff AL. Our estimates indicate that the highest exposures to benzene oc- curred in Akron I during the 1940s. B3ore the.l9JOs, the shorter work week (about 36 h) mitigated the h!gh workplace exposures and kept the weekly absorbed doserelativelj? lo&. After ihe 1940s, the improved engi- neering controls and the increasing concerns about the toxicity of ben- '?zene (as reflected in the decreasin TLVs) reduced the workplace con- centrations. The lack of significant P iofilm production in St. Marys in the 1940s had the effect of greatly reducing the expo>Jres of these workers (compared with Akron I ) during this period of time. The workers with the highest potential for exposure were those in the uen- P-len exposure-e and neutralizer exp\.\olosrukrine,@can.teegaor'trhiees.q'-Tuh.den6c-hf:edrsaiilny eaujva- e\\ron facilities w a 5 estimated to exceed 200 ppm lor most ot t e 1930s and -early m s . in the Akron racilities, we estimated that the ncut4izer op- -erators may have been exposed to concentrations in exc'kss nf 150 p p m for much of the same period. About 50 ppm equivalents or one-third a' the total -os ure to the n e wIiZpr \v;15 dLie to rlpr-ake. Our exposure estimates systematically incorporate contributions due to peak airborne concentrations and dermal contact into a single equiva- lent T W A airborne concentration of benzene. Table 1 4 presents the con- tribution of the three sources of exposure to the final estimates. On average, benzene concentrations due to fugitive emissions (background) contributed 5 0 4 0 % of the total exposure, depending on location. Peak exposures represented approximately 20% and dermal uptake consti- tuted 30% of the dose. The relative contributions from the three sources varied greatly over the different job categories and ior different years. The contribution of peak exposure to the workers' total exposure for certain job classifications such as mixer w a s less than 20'36, while other jobs such as the reactor, neutralizer, and quencher were doubled as a result of peak exposures. The contribution to total exposure f r o h dermal contact was high for hvo categories: quencher and neutralizer. The quencher job classification was estimated to have had the highest poten- tial for dermal uptake. Workers in the category received a dermal dose o i 14.5 mglkgld, which is equivalent to that received during 8 h of inhala- tion exposure to 65 ppm. Our estimates indicate that, on average, exposures for the different job categories at the Akron facilities were from one to eight times higher than the corresponding categories at St. Marys. There are two reasons for this difference. First, the finding that the St. h4ar).s facility was operat- ing at oiily 4 7 % of i t s capacity during the w a r years, 1912-1945, greatly reduced our estimate of exposure during this period (yet this is when levels at Akron I are believed to be at their highest). Second, the use of the A k r o n I I survey data to predict exposure a! the Akron 11 facility (rather than to the St. Marys data) results in higher estimates of wnrk- place concentrations for the years this facility operated (1938-1966). BESZEh't tXPOSURE OF PLIOFIU.( COHORT 219 The sensitivity analysis (Table 14) shoiced that no single factor p r o duced a major change in the overall estimate of exposure. For example, the assumption of high workplace concentrations during the early years of production had the greatest effect of all the factors in raising the estimates of exposures by 20%. The assumptions of an increased work week, analytical biases, use of personnel protection devices, and uncertainty in dermal estimates each changed the average estimates by less than 15%. Estimates for the years 1940-1949 presented similar results. The assumption of increasing workplace concentrations during the early years ofproduction raised estimates of exposures by 30%. All other factors again affected the estimates by less than 15%.The role of erroneous analytical measurement had only a minor impact on the estimates of exposure at either location. Our assumption that respirators only had a protection factor of 4 and were used only 33% of the time that workers were iniGolved in tasks where short-term concentrations exceeded the TLU proved not to have a dramatic effect on our estimates. For example, even i f it i s assumed that workers always wore their respirator when doing tasks with high exposure, and even if the respirators were 75% effective, our dose estimates drop by 10% in the Akron facilities and 9 % at St. M a r y s . However, if we assume that workers never wore respirators or that they probably of- fered little or no protection, our estimates would be only about 9% too low at either location. Lastly, dermal contact had contributed from 9-21 96 of the total average estimates, but for some exposure categories, dermal uptake results in doses twice as high as predicted b\(air sampling alone. 5 DISCUSSION Some of the methods used to conduct retrospective exposure assessments can be questioned, and it i s true that the results contain a signifi- cant degree of uncertainty. As discussed previously, only a handful of workplace measurements of benzene vapor are available for any of the plants for the years before the 1950s (Fig. 3). In the absence of analytical measurements, indirect evidence of exposure must be used. Such information includes later industrial hygiene surveys, production information, plant records, and worker intervieii;s. \$'e are a\vare of the potential pitfalls of relying on worker meniories in reconstructing exposure; however, in the absence of adequate written records, we believe that if recollections of different workers are relatively consistent, they provide a valid source of information. For example, the four employees interviewed for the paper had worked from 25-30 yr at the plant. The uncertainty analysis indicated that while no single assumption dominated the estimates, the factor that had the highest effect w a s the assumption of increasing workplace concentrations as one went backward from 1976-1940. Our belief that exposures during the early years of 2 2 0 D. 1. PAUSTE'JBACH E l AL. Pliofilm manufacture (pre-1950)were relatively high i s based on four factors: (1)medical and clinical evidence of toxicity in Akron and St. Marys workers (Department of Labor, 1942; Kipen et al., 1989a; Lamm et al., 1989; Wilson, 1942);(2) evidence of signiiicant engineering changes in St. M a r y s in 1946 and at Akron I in 1946; (3) industrial hygiene experience in other industries (Table 9); and (4) the higher TLVs that were in place in the 1940s and 1950s. We agree with Crunip and Allen (1984) that the commitment of rnan- agement to control workplace exposure to benzene to meet the prevailing TLV can be used to predict airborne levels before 1960. (:rump and Allen (1984) relied on a logical assumption that workplace exposure would be expected to have dccrcbased as t h e occupational exposure limi t s (e.g., the TLVs) decreased. Their estimztes are consistent with t h e reports in the industrial h)fgiene literature, for example, that workplace concentrations of benzene vapors in most industries declined between 1939 and the 1960s. Our analysis provides significant additional evidence to support Crump and Allen's contention that worker exposures to benzene were higher in the 194& and 1950s than in the 1960s and 1970s. The second most important factor, according to the uncertainty analysis, was the length of the workweek. The length of work week w a s based on records of personnel injury reports. Our estimates, while greater than 40 h, are less than the estimates from worker interviews. Workers consistently indicated that 6-and 7 d workweeks were the norm for all years after the 1930s. Had such estimates been include(', exposure estimates would have been increased by 5%. Although i t i s p,)ssible that the database i s not robust and may not be representative of all workers, the data seem reasonable in light of what i s known of other industries for that time period. Correction of analytical biases (in measurements taken before 1963) have a smaller effect on the final estimates then would hav> been expected. The reason for this lack of impact i s our use of the St. Marys data taken after 1963 (and thus unaffected by the analytical measurement bias) to estimate early exposures at St. Marys and Akron 1. While there were some measurements taken at St. Marys and Akron I before 1963, the data were so scattered that they were only used to confirm the estimates based on the post-1963 data (see Table 6). The impact of the analytical bias assumptions i s limited to the exposure estimated at the Akron II facility, where estimates of background airborne concentrations were increased by 50%. The contribution to total estimated exposure at the Akron I I facility was less than 50% since the contribution from tlermat contact was unaffected by the analytical bias. Both Crump and Allen (1984) and Rinsky et al. (1986)a;sumed that the available industrial hygiene data reflected the timcweighted average (8 h) daily exposure, including peak exposures. Crump and Allen (19&1) did attempt to incorporate peak exposures for workers who entered the dry- BEKZENE EXPOSURE OF PLlOFlLM COHORT 22 1 ers and casting units. In contrast, Rinsky et al. (1986) concluded that short-term, high-level exposures (even if they did exist) would not con- tribute significantly to the workers' total uptake (absorbed dose of ben- zene) because of the alleged use of personal protective equipment (PPEtgloves and respirators. In our analysis, we assumed thai respirators used during the years 1936-1976 had only a limited impact on exposure because of its limited effectivenessand inconsistent usage. We estimated that the half-face res- pirators used by the Pliofilm workers were no greater t h a n 75% effective. This assumption is based on the following lines of reasoning. First, it was recognized in the 1970s t h a t o n l y under ideal conditions \vould half-mask respirators provide a f i t that ensures that more t h a n 80% of the inhaled vapors pass through t h e charcoal (20% enters uncleaned through the leaks in the facial skinlrubber gasket interface). Based on the experience of the industrial hygiene community in implementing f i t tests during the 1970s and 19805, we would expect t h a t the respirators of the 1930s and 1950s would not be able to reduce exposure concentrations to levels beloLv 25% of the ambient levels due to the poor fit of t h e mask. Second, the records indicate t h a t these workers used Mine Safety Appliance (MSA) "comfo" half-mask respirators equipped with a char- coal cartridge (Mclnerney, 1977). Prior to 1975, charcoal cartridges were thought to effectively collect nearly all organic vapors. However, begin- ning with the classic studies of Nelson et 31. (Nelson and Harder, 1372, 1974, 1975; Nelson and Hodgkins, 1972; Nelson et al., 1972, 1974, 1975; Ruch et al., 19721, it was recognized t h a t at concentrations of 1000 ppm benzene, the time for 10% breakthrough occurs after about 2 h of use (Nelson and Harder, 1976). It also was recognized in the 1970s that for charcoal respirators to be effective they needed to be stored properly prior to and after initial use; othenvise the charcoal would soon be ex- hausted due to passive uptake of airborne organics (NIOSH, 1976). Worker interviews indicate t h a t it was common practice for workers to wear their masks around their necks when not in use; therefore, d u e to passive uptake and humidity effects, i t is unclear how effective the char- coal was at remcving benzene vapor (Allman, 1991). This is important considering the poor warning properties of benzene. Third, respirators offer only a modest level of protection for chemi- cals that have a n odor threshold greater t h a n or equal to the occupa- tional exposure limit. Such a warning property is necessary; otherwise, the worker could be exposed to harmful levels ulhile believing himself protected by the respirator. In the case of benzene, the odor threshold reported in the literature varies from 8 to 31 ppm (ASTM, 19/ti), 40 ppm (American Petroleum institute, 1985), and 100 ppm (Gerarde, 1963). Most hygienists believe it is a b 2 5 pom.fclrmn_<tneoDle a nd about 50 e p m 2for those . However, since 1958, the YLV has been 25 ppm (a value less than most estimates of the odor ' 2 2 2 D.I. PAUSTENBACH ET AL. threshold). It is also possible that workers would not have recognized that the respirators were not protecting them until concentrations within t h e respirator exceeded 100 ppm. It has b e e n observed that d u e to cen- tral nervous system depression a n d t h e onset of olfactory fatigue, e m - ployees have been unable to detect benzene`s characteristic o d o r at con- centrations as high as 200 ppm. In addition, smoking (whict was at its highest prevalence in US. males in t h e 1940s-1970s) can significantly raise t h e threshold of smell for b e n z e n e a n d other chemicals. Based o n the available evidence, even if o n e were t o accept that workers diligently wore respirators, it is likely that they offered only modest protection. T h e record on respirator usage by t h e Pliofilm workers is mixed. Rinsky et at. (1961) indicated that respirators were w o r n whenever t h e TLV w a s exceeded. However, such a policy must have been difficult to implement in light of t h e anal\.tical shortcomings during much of t h e 1930s and 1950s and t h e lack of a clear olfactory warning. In contrast, worker interviews indicated that while respirators were available and workers w e r e expected t o use them, they rarely w o r e them. This is n o surprise since benzene i s not an irritant and respirators at that time were most uncomfortable. As in many industries, workers often wore respira- tors around their neck but rarely put t h e m on (Allman, 1991. Mclnerny (1977 testified at the OSHA hearing on the benzene standard that respi- rator usage ranged from 25% in t h e 1930s a n d 1950s to 60% in t h e 1970s. Given this mixed testimony and the experience of industrial hygienists in other industries, we assumed that workers used respirators approxi- mately 33% of the time they were exposed t o peak concentrations. Although some of our estimates of t h e likely TLVA airborne concen- trations (150-250 ppm) appear high by today's standards, a n rsxamination of t h e ?Z?upational medicine literature (Table 15) indcatel that expo- sures t o these and higher levels can b e tolerated for variou-; lengths of time. inJhP For. l` e.x.am ! r uC, ple nf , workp 3ofl--\\'a< l a ce T: exposure to b e n z e n e for peric ds o m h mt ,I incornrnon ormr to 19-?hyand qxposure to levels from 100 to 200 pprn (8-h TW-\) could he tolerated for weeks by some persons without apparent significant adverse efiects (Hamilton, 1929; Aksoy, 1980). Greenburg (1926) studied people repeatedly exposed to benzene at 18 different work areas in a variety of facilities and s h o w e d that humans, if acclimated, could tolerate levels of exposure ranging from 1 ~ - 1 0 0 0 ppm. Table 16 presents s o m e of t h e results of this study. Concentrations in work areas in which benzene was the only solvent present ranged from 0 t o more than 4oOO p p m . The values reported by Greenburg (1926) represent time-weighted averages over several hours, and t h e concentrations given a r e "representative of general room air and/or air at a worker`s station." Elkins (1950) found that exposure to levels ranging from 200 to 700 p p m over several years resulted in s o m e fatalities (cause of death not stated) in t h e leather processing industry, BENZENE EXPOSURE OF PlIOFllht COHORT 223 TABLE IS. Acute Adverse Effects Associated with Shod-Term Fxpnsurc to Benzene Vapor Benzene Duration of concentration exposure (PPm) (h) 25 38 50-150 500 1500 1500-3000 1550-3100 1570- 3130 3000 m 3ccG-47Do a Unknown 5 1 1 Several hours 6 Several hours I/? I&-1 1 3130-4700 , k - 1 4650 I/? 4700 6190-9300 6?M-9300 7500 '0 Few hours I/?-1 Ih-1 6200-93M 2-4 19,m-20.w3 Shor: exposure Effects observed None Ollactory threshold Headache, lassitude, weariness Symptoms of illness Serious symptoms Moderate symptoms N o seiious effect Slight symptoms Endurable Dangerous Maximum concentration wlthout serious CNS effects Maximum concentration without serious Ch'S effects Listlessness, confusion, and some dizziness Confusion and some dizziness Definite symptoms of poisoning Immediate or subsrqucnt death Immediately dangerous to lile Loss of consciousncs5 Rapidly fatal Reference Cerarde (1960) API (1985) Cerarde (1960) Cerarde (1960) Cerarde (1960) Bloomfield (7951) Cirrberf (1939) Drowning (1937) Cerarde (1960) Browning (1937) Bloomfield (1951); von Oettingen (19401 Browning (1937) Hamilton (1929) Greenburg (1926) Greenburg (1926) Herbert (1939) Bloonilield (1951); vori Oettingen (1940); Gerarde (1960) Hamilton (1926) Erowning (1937l; Bloomfieid (1951); yon 0 c r i l n ~ ; c ~(1n9qoJ. Ceraidr (1W) although dermal exposure in these cases was probabl), significant. These findings are not surprising given the clear differences in susceptibility of the hematopoietic system among persons exposed to benzene. Individual case reports of acute benzene intoxication have appeared in the literature since the early 1900s (Greenburg, 1926; Hamilton, 1929; S a p pington, 1950). In certain studies, concentrations either were measured or estimated (Cook, 1947). Table 15 presents some of the published case reports which contain exposure information. Due to the likely underreporting by the analytical methods, much higher ,concentrations than those reported may have been present. The data indicate that concentrations of benzene up to 1000 pprn are tolerable for short periods and that concentrations between 250 and 700 ppm are not self-limiting i f L\,nrkprs are ac~c-limated to the smell. The data also indicate that airborne concentrations of benzene in the r a k e or 200-250 pprn (H-17 I W A ) may nbt necessarily cause obwous acute or subchronic toxicity. Often, in the "L 214 D.1. PAUSTENBACH E l AL. 1940s and 1950s there w a s a "selection process" where only those individuals who could tolerate the acute effects of benzene exposure remained on the job. For these reasons, we believe that o u r exposure estimates for the 1940s are not unreasonable. The finding that Pliofilm was not manufactured at St. Marys during 1942-1945, and therefore kvorker exposure would be less, raises a potential contradiction with the findings of Kipen et al. (1989a) regarding the effects on blood. In the Kipen et al. study, t h e worker's blood counts show some indication of an improvement in blood counts in 1943; however, they indicate a significant suppression in blood counts. Further, in I946 when rubber again became available and St. M a r y 5 returned to full production, blood coun!s showed no sign of renewed suppressim. A TABLE 16. Incidence of suppression' of LVhire Blood Cell Counts in 'v'arious lobs a s Compared to Range of Workplace Concentrations of Benzene (CreenbLrg, l z 6 ) - Benzene concentrationb Process \'e ni I I J t ion (P?m) Average Min Max 1ncider.ce of suppression Cement mixing' SlicAering' Coal Ins' h<ixing' Cement mixing' insulating wire' Compound mixing' Coating' Compound mixing' Core painting lire maklng Lining Compound mixing Cemeni mixing Lining Dry cleaning Co a i i ng Cementing None General room Local exhaust Enclosed process None None None General local exhaus! Enclosed process Sone None Local ekhaust None h'one Local exhaust None Local exhaust None 150 7w 500' 430' 150 130 210d 1360 583' 130 330' 1 22 0 150 2lOd io 9Od 340 620 le0 400' loo0 m 1M 100 5m 180" 470d 1Dc) 50 40' 84 2?O* 30 130d 100 113 140 M 50 0' 2w 310 20 2eO' 2 30 40 60 190 890 1029' 450d 190 210 464' 2QO 410 4CO' 100 340 160 340 110 350' 390 8io 360 SOY 4140 130 120 011 012 1/ 4 1I3 Oil 6112 711 Til0---- 113 2'5 119 0 '0 0'1 6!9 0;s 23 1/1 29 'Suppression i s defined as a 25% decreare in the individual's whl!e blood count. bMeasurement of airborne benzene wvere;ollecled wirhin the worL.rr5' brcathing zone during summer months. ' M r k e r s were exposed to olher solvents. dMearuremenrr were taken in the winter months. BEKZENE EXPOSURE OF PLIOFIU4 COHORT 225 plausible reason for this absence of a clear production-related improve- ment may be the way records were kept at St. Marys. Assignment to a job category was a function of the machine the worker operated, not the product manufactured or t h e solvents used. Therefore, a caster operator who manufactured tent fabric or foil products was identified as a caster o p erat or. Since blood measurements were both invasive and expensive, it can be expected that they were performed only on workers who were exposed to benzene (Hanning, 1991b). Consequently, it seems likely that the blood data for St. Marys during the war are only for those employees who were involved in the reworking of Pliofilm. Some indication of this can be seen in the decline in t h e number of workers tested in the war years. For example, only 6 workers' blood counts were reported for a 17- month period starting in February of 1942 (Kipen, 1992). Such workers would have had exposures greater t h a n the estimates developed in Table 13. The estimates in this table assume that all workers at St. Marys had an equal probability of manufacturing Pliofilm during the war years. A second possible explanation is that monitoring of blood continued during the war for those workers who had previously developed dyscrasias. It is known t h a t some workers exposed to benzene never returned to preex- posure levels and yet remain asymptomatic. During our investigation, we identified a significant difference be- tween the medical monitoring programs of the two Akron facilities and St. hlarys. Red and white blood cell counts were performed on a monthly basis at St. Marys from the 1930s to 1976 (Department of Labor, 7942; Hanning, 1930; Rinsky et al., 1981). In A k r o n , the blood monitoring, initially was conducted each month, but w a s relaxed to quarterly about 1913. The impact of this decreased monitoring frequency is that a worker could continue to be exposed to benzene for nearly 3 m o before a blood dyscrasia could be identified. At St. Mavs, the worker could have continued to be exposed for only 1 mo, t h u s reducing the possibility of devel- oping serious blood dyscrasias. Beyond the evidence for higher airborne concentrations, w e believe that this may help explain the apparent in- crease i n leukemias observed at the Akron facilities. For a number of reasons, some caution should be used in interpret- ing the data from the Pliofilm cohort. The first is the efiect of "selecting for tolerance." It appears that there can be significant differences in interindividual susceptibility to the effects of benzene. As such, it i s u n clear whether the information on this cohort is biased since persons susceptible to the hematopoietic effects of benzene may have left the workplace prematurely as a result of the program of systematic testing and removal of workers. Whether this selection for tolerance would bias the cohort`s response to the lukemogenic effect of benzene is not known. A second uncertainty is that some workers may have had additional exposure to benzene d u e to work in other settings. For example, 22 b D.1. PAUSTENEACH E l AL. at the Akron I facility, benzene was used in the same building and on the same floor as the Piiofilm operations in the manufacture of balloon fabric (Department of Labor, 1942; Hanning, 1991; Krisman, 1991). It i s known that some workers in the Akron facilities were transferred to Pliofilm from other departments where benzene was in use. In contrast, the St. M a r y s Pliofilm operation was the only department that used significant amounts of benzene. This suggests that the lifetime estimates o f benzene exposure for the Akron workers may not be fully understood. Although this analysis is based primarily on exposure data which were considered in the OSHA hearings on the benzene workplace stan- dard, it i s inaccurate to suggest that all of the pertinent data were consid- ered or that they were properly integrated or interpreted (Infante, 1992). As discussed throughout this article, no prior assessment has attempted t o quantitatively account for t h e five important exposure factors we ad- dressed. Further, some cited information, such as the Lirtual shutdown of production of Pliofilm at St. M a r y s during the war and the exceedingly long workweek in Akron and St. May., were not considered by OSHA. Last, w e believe that the blood data and the acute toxicity observed in some workers are inconsistent with t h e exposures suggested in the analysis of Rinsky et al. (1986). Like OSHA, we agree that the Crump and Allen ratio approach is likely to be the best way to retrospectively estimate dose and have adopted this a proach in our assessment. Although additional data regaring the exposure of the Pliofilm co- hort would give us greater confidence in our analysis, we believe that our estimates are more plausible than those of earlier studies. Our results predict exposures that are much like those in other industries at that time and are consistent with those known to cause blood suppres- sion. f3y considering the differences between the iacilities at the two locations, we determined that prior studies probably overesti~atedexposures at St. Mary and underes!irnated exposures at Akron. We also concluded that dermal uptake M'X likely to contribute at least an additional 20% to the dose absorbed via inhalation. We suggest that these exposure estimates for the various classes of workers b e compared with the incidence of disease information to see if there is a doseresponse relationship. If it i s present, we recommend that these estimates be com- bined with t h e most recent epidemiology data on the Pliofilm workers to kderive a cancer potency factor for humans. LVe also sug lest that when additional or better information on the Pliofilm workers ecomes available, it be incorporated into the analysis. REFERENCES Agency for Toxic Substances a n d Disease Regist') 148Y Toxicologral Prohl? for Benzene. Atlanta Ca: U.S. Public Health Service. BENZENE EXPOSURE OF PLlOFllhl COHORT 217 Aksoy, M. 1980. Different types of malignancies due to occupational exposure to benzene: A review of recent observations in Turkey. Environ. Res. 23:181-190. Albert, R. E. 1979. Carcinogen Assessment Group's Final Report on Population Risk IOAmbient Benzene Exposures. U.S. Environmental Protection Agency, Officeof Air Quality and Planning Standards. E P A - 4 W W . Research Triangle Park, N.C. Allman, C.1991. Personal communication wilh R. Bradshaw, ChemRisk, and W. Ollison, American Petroleum Institute, January 3. American Conference of Governmental Industrial Hygienists. 1991.Notice of intended changes- Benzene. Appf. Occup. Environ. H y s . 5(7):453-463. American Petroleum Institute. 1985. Review of Published Odor and Tasfe Threshold Values ofSolu- ble Gasoline Componenfs. Health and Environmental Sciences Depanment. API publication no. 6119. December. American Petroleum Institute. 1986.OSHA Hearing Docket HOj9c. Exhibition 216, Tabies 12 and 15. Washington, D.C. American Society for Testing and Materials. 1978.Compilation oi Odor and Tasfe Threshold W u e s Dara. ed. F. A. Fazzslari, p. 17. Philadelphia: American Socieiy for Testing and Materials. Austin, H., Delzell, E., and Cole, P. 1988. Benzene and leukemia: A review of the literature and a risk assessment. Am. /. Zpidemiol. 127:419439., Beebe, P. 1976.Letter to J. A. Brown, Diredor of Fitm, Flooring and Molded Products Manufacluring from P. Beebe, Ir.. Manager Industrial Hygiene and Safety. Re: Pliofilm Operations at SI. Marys. September 20. Bernardinelli, S. P., and Bender. T. R. 1990.Selection of chemical protective clothing. I n xogenous Derma:oses: tnvrronmenlal Dermarifrs, chapter 36. eds. T. Menne and H. I. htaibach. Boca Raton, Fla.: CRC Press. Bloomfield. C. D. 1951.Studies of health hazards i n industry. Chemical hazards: Benzol vapor. Ind. Health Monfhfy 11:62-64. Bond, C. 1991. Survey ol \'arrous Concenfraf/onr of Benzene hfeasured During b'aarious Chemical h!anuf'acfuring Operalions Befween 1944-1965. Midland, Mich.: Dow Chemical Company. Brett. S. At., Rodricks, J . V., and Chincilli. V. M. 1989. Review and update of leukemia risk potentially associa:ed with occupational exposure to benzene. nviron. Heafrh Perspec(. 82:267-281. Browning. E. 7937. Toxicity ol Industrial Organic Solvents: Summaries of Published U b r k . Report No. 00.Committee o n the Toxicity o f Industrial Solvents. Medical Research Council, Industrial Health Research Board. London. Cody, R. P.. Strawderman, 1'4. b'.,and Kipen, H. M. 1991. Hematologic effects of benzene: Job specific trends during the first year of employment arnoiig a cohon of benzene-exposed rubber workers. i n preparation. Cook, LV. 1915. Maximum allowable concentrations of industrial contaminaats. Br. 1. lnd. M e d . 14(11):93&-916. Cook. 1%'A. 1947. Review of aulomJttC indicating and recording instruments f o r determination o f industrial atmospheric con!aminan[s. Am. Ind. l f y g Arsoc. Q. 8:4?-43. Crisman, H. 1991. Personal Communication with D. Paustenbach, ChemRisk-A Division of McLaren/ Hart Engineering, Alarneda, Calif., January 3. Crisman, H. 1991. Personal communication w i t h D. Paustenbach, ChemRisk, December 27. Crump, K.. and Allen, 8. 1984. Quantitative Estimates of Risk of Leukemia from Occupational E x p o s u r e t o Benzene. OSHA Docket H459b. Exhibit 152 (Appendix B). Washington, D.C. Depanment of Labor. 1932. Condensed Proceedings of the Conference o n Health Hazards in the R u b t x r Industry U.S. Depanrnen! of Labor, Division of Labor Slandards. Akron, Ohio. May 29. Dorsey, C. P. 1973.Industrial Hygiene Report of Benzene Samples Collected During Various Phar- maceutical Operations. Eli Lilly and Company, Indianapolis, Indiana. Drager. 1980. The Handbook lor Dra'ger Tubes: Air Inverfigafions and Techn/caf C a s Analysis wifh DrJger Tubes, 3rd ed. (August 1976).N e w York: Drager. Elkins, H. B. 1950. The Chernisfry of Indusfrraf Toxicofogy. pp. 101-107. New York: J o h n L\'iley and Sons. 228 D. 1. PAUSTENBACH ET AL. Flanick, H . L. 1943. Memorandum to H. L. Flanick from H . Slomin. Re: Employment and Labor Requirements in the Rubber Manufacturing Industry, June 21. Fluker, 1. H. 1%. Letter to J . H . Fluker, Assistant Superintendent, Division of Safety and Hygiene from \.2! E. Obetz, Medical Investigator for T h e Industrial Cornmission o f Ohio. Re: Site vis11 to Coodyear Tire and Rubber Company, St. Marys, O h i o to check the effectiveness ofthe ne* ventilation equipment. November 21. Fluker, J. H. 1 9 4 Letter dated M a y 4. Re: Results of an industrial hygiene survey on The Cocdyea: X r e & Rubber Co. Plant, Akron, Ohio. Foreman's Personal Injury Reports. 1910-1956.St. M a r y s , Ohio. Obtained via h o d y e a r ' s worker history files. Franz. T. J. 1984. Percutaneous absorption of benzene. In Applied Toxicology olPefroleum Hyc*o- carbons, ed. MacFarland, H. N.. Holdsworth, C. E., hlaccregor. I . A.. Call, R. VU',, and Lane, I.!. t., p p . 51-70. Princeton, N.J.:Princeton Scientific. Cerarde, H.W. 1460. TOXiCO/Ogy and Biochemisrry ol Aromatic Hydrocarbons. London: Else\ ier Publishing. Cerarde, H it! 1%) The aromatic hydrocarbun, In / n d u s f r i a l ! l y & i c n e a n d Toxicolo&y,VIJI. 2 e 5 f . A. Patty. New York: john Wiley and Sons. Coldstein. B. D. 1977. Hematotoxicity i n humans. 1. Toxicol. inriron k l e a l t h (Suppl.) 2:69-10; Coldstein, 8. D. 1953. Benzene: Review of Recent Literature. Prepared for t h e American Petrolea-;i Institute. February 3. Greenburg. L. 1926. Benzol poisoning a s an industrial hazard L'l, Intensive study of selected initis. tries with respect to factor conditions and polluticn of the atmosphere by benzol. Pun`ic Healrh f i e p 41:lSlb-1539. Hamilton, A 1929. Induslrial Poisons in f h e United S f a r c : . pp. 453-481 h e w York: h l ~ c m i l l a : ~ Hamilton. A. 1931. Benzene (benzol) poisoning. Arch. Patho:. 11.134-601. Eankc. 1.. Duthiewicz, 1..and Piotrowshi. J. 1961. The absorption of benzene lhrough In(. s k i n i r , man. h f e d Pracr. 12 413-426. Hanning. R 1990. Personal comnwr\icaiton with R. BIJd5hJW. ChenrRisi, and W. Ollison. American Petroleum Institute Hanning. R . 19912. Personal communication with D. Paustenbach, ChemRisk. and i','O.ilison, American Petroleum Institute, May 1. Hanning, R. 1991b Personal conimunicalion with t\! Ollison, American Petroleum Institute, lune 6 Hay. E. B.1W. Exposure to aromatic hydrocarbons in a coke oven by-product plant. Am Inc' Hjg ASSOC1. 25:326-391. Hay, E. B. 1990 Personal communication by kV. Ollison. Herbert. E. Q. 1939. Dangerous gases in distillation and refining processe;:TT/nsr Perrol. IS 323- 346. Hillman, S. 1942. Letter to 5. Hillman, Associate Director. General Of!Ace of Production hta-;;e. men1 from the Secretary o f Labor. Division of b b o r Standards, National Archives. :.g. ton, D.C.. January 10. Hornung. R. LV., Ward, E., Morris, J. A,, arid Rinsky, R. A. 1989. Letter 13 the editor. Toxicc! / n d H e a l r h 5:1153-1155. IARC. 1982. Discussion o n benzene, IARC Fval. Carcinogen. R i s k Chem. Hum., pp. 95-14: IARC Monograph no. 29. Lyon, France. Industrial Commission o f Ohio. 7%. Self.lnsuring Employer's Report of Occupational Disease. Columbus, Ohio. Obtained from worker records;.this i s Rinsky worker number 633. I n f a n t e , P. F. 1992. Benzene and leukemia: Ttic 0 1 ppm A C C l H proposed threshold limit va:sc. f o r benzene. Appl. Occup. Environ. Hyg. 7(4):253-262. Infante, P. F., Rinsky, R. A.. Waggoner, L. R.. and Young, R. 1. 1977. Leukemia in benzene wafj.eri. lancer 2:7b-70. Infante. P. F., and \j'hite, LV. C. 1903. Benzene: Epidemiologic observations of leukemia ce:l t)pe and adverse health effects associated with low-tevel exposure. Environ. Health Perspecf 52175-82. BENZENE EXPOSURE OF PLlOFlLM COHORT 229 Kipen, H. A i . 1992. Personal communication with I! Pricc. ChemRisL-A Division ol McLarenlHJrt Engineering. Portland, hiaine. January, Kipen. H. ht., Cody. R. P., Crurnp. K. 5.. Allen, D. C.. and Coldstein, 8.D. 1908 Hcmatologic effects of benzene: A thirty-five year longitudinal study of rubber workers. Toxrcol Ind. Healfh 4.411430. Kipen. H. M., Cody, R . P., and Coldslein, 8. 0. 1989a. Use 01 longitudinal analysis of peripheral b l d counts to validate historical reconstruction 01 benzene exposure. Cnviron. Healrh Per- specf. 82:199-206. Kipen. H. hl., Cody, R. P.. and Coldstein, B. D. 1989b. Letter to the editor. Toxicol. Ind. Heahh 5.1156-7 158. Kodah. 1991. h l e m o to Dennis Paustenbach, ChernRisk-A Division of h4cLarenlHarl Engineering, Alameda, Calif. Kusnetz, H. I.S,altzman. B . E,, and Lanier, M. E. 1960. Calibration and evaluation 01 gas detecting tuber. Am. Ind. Hyg Assoc. 1. 21361-373. L m m . 5. H., Walters, A. 5.. Wilson, R.. Burd, D. M., and Crunwald. H. 1989. Consistencies and inconsistencies underlying the quantitative assessment of leukemia risk from benzene exposure. fnviron. H e a l r h Perspecl. 02:289-297. htaibach, H. I.. and Anjo. 0 . M. 1981. Percutaneous penetration o f benzene and benzene con- ta:ned in solvents used i n the rubber industry Arch Envrron. Heallh 36;51.?56-260. htclnerney. H. hi. 1977. 'Testimonyof July28.Submitted to OSHA Hearing Dochet H059. Exhibit 121. N IOS HI\<'as hinp,t on, D.C. National Institute for Occupational SJfety and Health. 1976. A Curde for Indusrrial Respiraiory Proiecfton john A . Pritchard, 10s Alamos Scienrilic Laboralor).. u n l e r interagency agreement nos. IA.71-?3. IA-75-25, and IA-7&9 lor U.S. Department of Health. Education. and LUeIfare, Public H e a l t h Service. Center lor Disease Control and National Institute for Occupational S a f e ? and Health. Cincinnati, Ohio: NIOSH. h'elson, C. 0.. Correia. A. A I , , and Harder, C. A. 1975. Respirator Cartridge Efiiciency Studies VII. Efiect of Relative Humidity and Temperature Lawrenre Livermorc Labor.>tory, report UCRL- 7733.3. Aupsst Nelson. C. 0..and Harder, C. A. 1972. Respirator cartridge elficiency studies. lit.Elfecis of steady s t a t e and pulsating llou.. Am. Ind H),g. A S S O CI. 33 797 Nelson. C . O., and Harder, C A . 1974. Respirator cartridge elliric.ncy stt~dier.V. T f k c t 01 sol~~.1'ii1 vapor. Am. Ind. H1.g AJSOC.1. July. 391-410. Nelson. C. 0..and Harder, C. A. 1976 Respirator cartridge efficiency siudles \'I. Effect of concen- tra:ion. Am. Ind. Hyg. Assoc. /. April: 205-216. Nelson, C. 0..Harder, C.A.. and 6igler. E. E . 1974. Respirator Canridge E f t , c i e n c y Studies \'I. E f f e c t 01 Concenlration. L a h r e n c e Livermore Laboratory, repor1 UCRL-~LIS;.Nowmber. Kelson. C. 0..and Hodgkins. D J. 1972 Respirator cartridge effictency srudles I1 Prtrparaiion o f t e s t a!mospheres A m Ind Hyg A S S O C1.. 33:llO. h'elson, C 0.. Johnson. t . R.. and Lindeken, C. L., and 'Taylor. R . D 1972 Respirator cartridge e l f i c i e n c ) siudies Ill. A mechanical breathing machine to simulate human respiration. Ani Ind. Hyg. ASSOC. /. 33:745. &elson. C 0..Lum. B. Y., Carlson, C. I., tl'ong. C. hl.. and Johnson, 1, 5. 1981. Glo\*epermeation by organic solvents. Am. Ind Hyg Assoc. 1. 42:217-225. Nomiyama. K . , and Norniyama, H. 1974. Respiratory retention, uptake, and excretion of organic solvents in man. Benzene, toluene, mhexane, trichloroethylene. acelone, ethyl acetate, and ethyl alcohol. In[. Arch. Arberismed. 3275-63. (As cited i n IARC. 1982.1 Occupational Safety and Health Administration. 1985. Occupational eiposure to benzent.; P r o posed rule and notice of hearing. Fed. Reg. 52:50512-50586. Occupational Safety and Health Administration, 1987. Occupational exposure to benzene; Final rule. Fed. Reg. 5?:36360-U578. OSho:ne, C. 1931. Personal communication with R. BradrhJw, ClicmKi$L. arrd it' Ollison, Amcri- c a n Petroleum Institute. January 3. . Oit#hl. C.,To\*nsend. I . C.. Fishbeck, w. A., and Largner. R A. 15-8 ~ o n ~aml o~ngvInd,vlc;ua~, occupationally exposed to benzene. Arch. Envtron. Hea!rh 3 3 . 3 - 1 0 , Pagnono, L. D.,Elkins. H. E., Brugsch, G..and Walkley. I . E. l % t . lndurrrial benzene ,.xw,urc iron) petroleum naphtha I . Rubber coating industry. Am. /nd. ~~8 A~~~~ 1. ~ : 2 1 7 - ~ 4 5 , Pauslenbach, D. I.1977. Industrial Hygiene Report 01 Benzene Sampler ColiK1cd daring vd,lo-s Chemical Manufacturing Activities. Eli Lilly and Company, C\in!on, ~ n d , Panon, &{. E., Chinchilli, V. hl., Bret:, 5. M.. and Rodricks. I. V. 1992. Reanalyrls and of leukemogenic risk associafed with occupationaf benzene exposure i n the pllofllrn R i s k Analysis, submitted. ginsky, R. A , Smith, A. B., Hornug, R.. Filloon, R. c.,Young. R. I., Okun. H. A,, and Landrigan, p. ti. 1986.Benzene and leukemia. An E p i d e m d o g i c Risk Assessment Cincinnati, ohlo: ~ a 1 ~ 0 n ~ i Institute for Occupational Safety and Health. Rinsky, R. A,. Smith, A. B., Hornug. R., Fil!oon. R. G.. Young. R . I., Okun. A. H.,and bndr,gan, p. 1 , 1967. Benzene and leukemia: An epidemiologic risk assessment. N. Engl, 1. ~~d 3 1 6 . 1 ~ - 1050. Rinsky, R. A,, Young. R. I., and Smith, A. B. 1931. Leukemia i n benzene workers. ~ m1..lnd, h j e j 2:217-235. Rothstein, D. E. 1942. Memorandum from L. Levine to D.E. Rothstein. W a r Production Board, p.e Development in the synthetic rubber industry. July 2. National Archives, LVashinpon, D.C. Ruch, \\I, E., Nelson, C. O., Lindckan, C. 1.. lohnson, R. E., and Hodgkins, D.J. 1972. Respirjrvr canridge efliciency studies I. Experimental design. A n . Ind. H j z I. 33:105. Sappington, C. 0.1950. Occupations( morbidity and monahty. In Essentials of Industrial Healrh Philadelphia. I . B. Lippincott. Schoch, D. 19'31.Survey of \brious Reported Concentrations of Benzene Measured near Different hianufacluring Steps. Rochester, N.Y.: Eastman Kodak. Sefarian, S. 1956 Repon on Investigation of Benzol fxposures within Coodyear T i e and Rubber Compan! SI. htarys. Ohio. January 17. Silverman, L. 1936. Sampling and analyzing air for contaminants i n work placer. In IncyclopeofJ o / / n s f r u r n e n : a ! l o n (or lnduslriai H),glene. Ann Arbor, h5i:h.: University o f h5ich:gan. TJbershJw, I. 8 . . and Larnm. S t i , 1977.Benzene and leukemia ( l e t t e r to the editoi) l a n c e r 2 867- M-8 Unltcd Rubber IVorkers 1977. United Rubber Worker Testimony. OSHA Hcaring Docket H 0 5 9 . OSHA Hcaring Transcript, Ausust 10. Washington, D.C. Unwc'rit) 01 h'orih Carolina 1971. Preliminary Environmental Survey Repcrt. S t . hlavs. Ohlo. dated Novt.mber 11, 1974. Exhitnt 187.C.6,DocLct HO59, Submitted I OOSHA dociel on twn, cI 13 ne. 0:c u pa!10 na I Sa f c: y a nd t4 ea I!h A d m inist r a 1ion, \%`a s hin[,t o n, D . . U.5. fnvironniental Protection Agency. 1985. Interim Quantitative Cenie7-iinit Risk Estimates dtie t o InhJl.ition ul D v n i r n c C)ifirr 131 Air Q u a l i t y Planning and Standards, Carcinogen AsseSj- merit Group. EI'A~6ClC)!X8j0!2. February 15. \'~gliarii, C C 19:b. Lrukcmia associated h i t h benzene. Ann. h'\: Acad. Scf. 271.143-151. \ o n Oertingr4rt,\':. F. 1940.Public Healrh Bull. 225. Washington, D.C.: U.S.Public Health Service .L'oy7c-L. I', I and Thorsiurid, T it' 1591.B r n i e n e risk assessment Status of qusntibing the leu;.+ mogenic risk associa!ed \\ith the low dose inhalation d benzene. Risk Anal. 11(31:35;-357. \VeeLs, I;.V, , a n d hicleod. hf. 1. 1982. Permeation ofprotective garment material by benzene and titrated wz!cr. A m . Ind. H } , g . Assoc. J. 43:ZOl-211. tYhitc, hi. C., Infante, P. F., and Chu, K. C. 1982. A quantitative estimate of leukemia mOrtJliv associated with occupational exposure to benzene. Risk Anal. 4:9-13. Wilson, R, 1942. Benzene poisoning in industry. 1. Lab Cltn. Med. 27:1517-1521. Wingloot Clan. 1983.Operations at SI. M a y s plant span 43 year period. Goodyear Tire and Rubber ConlpJn)' h'rb/Cff Sf. h4a9.5 fd 4?(4).1-5. Wong. 0 .1983.An Indurtn.\Vide Study of Chemical Workers Occupationally Exposed 10 Bcnzene. Report to Ihc Clwniical Manufacturers Association (ChlA), Environmental Health Associates, Berkeley, Calif. Yin, S.-N., Li, C .L.. lain, F.-L..Fu, Z .I., Jin, C., Chcn, Y..]., Luo, 5.1.. Ye, P.-Z., Zhang. J.-Z., M'and. C: -..- ---------;- BENZENE EXPOSURE OF PLIOFILh\ COHORT 231 c., Zhang. X.-C.. \Vu, H.-N.. and Zhong. Q .C. 198;. Leukemia in benzene workers. A relro. Speclive cohorr study I. General results Ut ). / n d , hfed 44.124-128. Young. K., Rinskey. R.. and lohnson. P. 1 9 7 . \V:JIL..Through Survey of Coodyear Tire a n d Rubher co. Cast Film Division. St h \ J r y c . Ohio. Survey dale February 9-10, 1976. Survey conducwd by P. lohnron. R . Rinrky, P. Intanre. hq. lonrs. and R. YounF trom the Division 01 Field Studies. Hazard Evaluations. and Field S ~ u d i e sN. ational lns\itute lor Occupational Safely and Health. Cincinnati. Ohto. zimmer. V. A. 1941. letter to 6 . hlcCurn from \', A. Zimmer. Director, Division of Labor Standards. Department of labor, December 20. Re. Labor shortage during World War 11. National Archives, W'ashington, D.C. Received july 75, 7491 Accepred /anvary 75, 1992 I I-- --- 1I