Mortality among workers engaged in the development or manufacture of styrene-based products--an update.

Mortality among workers engaged in the develop ment or manufacture of styrene-based products - an update. Scand J Work Environ Health 1992; 18:145-54. Mortality was updated another 11 years through 1986 for a previously studied cohort of 2904 male chemical workers who were potentially exposed to styrene and related materials for a year or more between 1937and 1971. Substantial deficits in mortality from all causes and total cancer were ob served in the cohort when it was compared with white males in the United States, and also other chemical workers who were unexposed to styrene-based products. Mortality from leukemia was slightly less than expected during the updated period, in contrast to an excessof lymphatic leukemia observed in the origi nal period. Yet small elevations in risk of other types of lymphatic cancer, particularly multiple myeloma, persisted. The risk of these cancers did not increase with estimated intensity or duration of styrene ex posure. The findings are discussed in context with those of studies of similarly exposed workers in related industries.

than one year in the work areas of interest were excluded . Twenty-five female employees were also excluded because of their small number .
The original study used a complex exposure classification scheme to handle both the multiple chemical agents in the work environment and the varying levels of exposure that existed between jobs. For consistency, this coding scheme was retained for the updated study. A brief summary of the method is presented below, but once again the reader is referred to the original report (1) for a detailed description.
Manufacturing jobs held between 1937 and 1977 were grouped accord ing to similar exposure profiles to form 57 categories of common exposure experiences. Each of the categories was individually evaluated by an industrial hygienist and was assigned an exposure intensity code with respect to the five chemical agent groupings displayed in table 1. Some of the categories were judged to have involved exposure to more than one of the agent groupings. To improve the clarity of presentation, the convention used by Ott et al to name the agent groups (ie, vapors I , vapors 2, etc) was abandoned in favor of explicitly present ing the primary agents and exposure levels as outlined in table 1.
Because of their idiosyncratic natu re, research jobs were judged as not classifiable with regard to exposure to each of the agent groupings . They were of epidemiologic interest anyway, so they were included in the study, but were analyzed separately.
Coded information on each worker (ie, employee name, employee identification number, social security number, job assignments coded to the 57 common exposure experience categories, and associated date s of assignments) was retrieved from the key-punch cards Scand J Work Environ llealth 1992, vol 18, no 3 used for the original study. A file of the cohort was then linked to the company's central mortality surveillance data base to identify both the vital status of the workers as of 1 January 1987 and the causes of death of the deceased. This mortality surveillance system has been established and maintained to permit the efficient conduct of health studies of the Dow work force (2). Table 2 presents the updated employment and vital status of the cohort. During the 11 years of additional follow-up, the total number of deaths among the workers increased from 320 to 687. (One additional death which occurred prior to 1976 and was missed in the original study was identified.) Nine workers (0.4%) could not be traced and were considered lost to the follow-up.
Person-years at risk were accumulated for each eligible worker across five-year age and calendar year intervals beginning with the qualifying date of exposure (ie, one year after first assignment in a styrene work area for the general analysis or 1 January 1940, whichever was later) and ending with the date last known alive, or 1 January 1987, whichever was earliest. The cohort members contributed nearly 90 000 person-years at risk for an average follow-up of 30.9 years per worker.
Cause-specificmortality among the cohort was compared against that of the white male population of the United States (US) with the use of a modified life-table program developed by Monson (3). The standardized mortality ratio (SMR) was calculated as a ratio of the observed to the expected number of deaths multiplied by 100. The 95070 confidence interval (95% CI) was calculated for each SMR estimate.
Analyses were done of mortality across major work areas (ie, styrene monomer and finishing; styrenebutadiene latex; product research and development; polymerization, coloring and extrusion), agent groupings, mutually exclusive groups defined by exposure and socioeconomic status, duration of exposure, and latency. Employees could contribute person-years at risk to several categories of those variables consistent with their time dependency. The trends across SMR estimates specific for different exposure groupings were estimated with the use of Breslow's chi-square test with one-degree of freedom (4).
Mortality among the cohort was also compared with that of all other Michigan-based Dow employees, as adjusted for age, interval since entry into follow-up, and pay status (ie, salary versus hourly). This comparison was achieved through the direct adjustment of mortality rates according to an adaptation of the Mantel-Haenszel method for cohort studies (5). The Michigan-based cohort was chosen for comparison purposes because more than 80% of the styrene-based cohort was also from the Michigan location. It includes a sufficiently large mix of administrative, production, and research staff, a small percentage of whom have had occupational exposures to other potentially haz- a Employees can be counted more than once since some could have worked for one or more years in multiple plants .

Results
ardous materials, including vinyl chloride, asbestos, chloromethyl ethers, and organic dyes (2). These comparison workers were more likely to be socioeconomically similar to the exposed group and to share similar life-style experiences than the US white male comparison group.
Overall, in comparison with the US population, the cohort continued to experience statistically significantly fewer than expected deaths from all causes combined, including nearly every major category (total cancer, arteriosclerotic heart disease, nonmalignant respiratory diseases, diseases of the digestive system, and external causes of death) ( There was a nonstatistically significant increase in mortality from cancers of lymphatic and hematopoietic tissue (SMR 144, 95% CI 95-208). This finding was due to small to modest excessesof mortality from each of the individual causes that comprise the category. Ott et al (1) originally noted seven deaths coded to leukemia and three additional deaths coded to other causes, but for which the death certificate mentioned leukemia as a contributing cause or under other significant conditions . They also discussed one living case of leukemia. During the updated follow-up period, two new deaths occurred from leukemia compared with nearly four expected deaths from this cause. These two included a megakaryocytic leukemia and a subacute granulocytic leukemia. In addition, the one living person identified by Ott et al as having leukemia eventually died in 1984 of a myocardial infarction, 13 years after having been first diagnosed with chronic lymphocytic leukemia.
During the updated period two additional deaths occurred from Hodgkin's disease, four from non-Hodgkin's lymphoma, and four from multiple myeloma. These values compare with 0.6, 3.8, and 1.8 expected deaths from these conditions, respectively. It should be noted that one of the new deaths from multiple myeloma was discussed as an incident case by Ott et al (1) .
The same general mortality patterns persisted when thecohortwas instead compared with allotherMichigan-based Dow employees (table 4). However, the excess multiple myeloma was, in this instance, statistically significant [rate ratio (RR) 2.45, 95% CI Analysis of mortality across the four major work areas (table 5) showed a nonstatistically significant excess of lymphatic and hematopoietic cancer among employees from the polymerization, colorin g and extrusion work area. In particular, multiple myeloma was elevated in this group. By contrast, the excess mortality in the categor y of symptoms, senility and ill-defined conditions occurred exclusi vely in the styrene monomer and finishing work area.
The mortality from nearly every cause of death analyzed was less than expected for the employees who worked in product research and development. This was true for both salaried professional and management personnel and hourly technicians and other support staff who worked in this area (not shown). There were, however, nonstatisticall y significant increases in mortality from Hodgkin's disease and non-Hodgkin's lymphoma, based on two cases of each, among the professionals .
An examination of mortality across the agent groupings taken individually (not shown) revealed deficits of mortality from all causes combined and total cancer within each group. There were statistically significant increases, however, in mortality from lymphatic and hematopoietic cancer among the workers categorized as exposed to styrene and ethylbenzene in the low range of a time-weighted average (TWA) of 1-4 ppm (SMR 236,95070 CI 122-411) and from arterio-sclerotic heart disease among the workers exposed to a TWA of~5 ppm of styrene and ethylbenzene (SMR 134,95% CI 104-171). Further investigation of the heart disease excess showed it to be concentrated primarily among the persons with short ( < 1 year) to moderate (1-4 years) duration of exposure . Few of the deaths from heart disease occurred while the employees were in exposure areas, the majority occurring only many years after the workers had left work in these areas. Table 6 presents a summary of mortality from lymphatic and hematopoietic cancer by duration of exposure to each of the agent groupings. (Only those groupings with at least two lymphatic and hematopoietic cancer deaths are shown.) None of the excess mortality in this category of death appeared to be concentrated in the longer-term exposure category. Instead, the excess appeared to be confined to workers who were exposed for less than five years. In addition , there was no apparent relation to intensity of styrene exposure, since the risk among workers in the high exposure group (ie,~5 ppm) was unremarkable. There was no pattern evident when lymphatic and hematopoietic cancer mortality was examined by year of first exposure. Analyses allowing for a 15-year minimum latency period revealed a statistically significant mortality risk (24 observed , SMR 160, 95% CI 102-238). However, there was no significant trend of increasing Table 5. Observed (0) and expected" (E) deaths and the standardized mortality rat io (SMR) by major work area -styrene-based products cohort, 1940-1986. b Work Area d f Statis tically sign i ficant at a =0.05, two ta iled. Table 6. Observed (0) and expected" (E) mortality from lymphatic and hematopoiet ic canc er by agent grouping and du rat ion of ex posure -styrene-based product coho rt , 1940-1986. Employees may be counted more t han once in the table since some co uld have been exposed to several agent groupings as a consequence of job mob ility. C Expo su re duration calculated through 1 January 1976. d Stat ist ically significant at a = 0.05, t wo-tailed. risk across three categories (ie, < 15, 15-34, and 35 years) of time since first exposure (P = 0.36). The subjects were allocated to one of 13 mutually exclusive groups on the basis of exposure and socioeconomic considerations (table 7). This grouping showed the elevated lymphatic and hematopoietic cancer mortality to be mostly confined to persons who were professionals (7 observed, SMR 177, 95070 CI 71-365), and to production workers who were exposed to extrusion fumes and directly or indirectly to colorants (9 observed, SMR 263, 95% CI 120-500).

Discussion
In 1980, Ott et al (1) reported the mortality experience of this cohort of workers through 1975. At that time, the group displayed fewer than expected deaths from all causes combined and from malignant neoplasms. An excess of lymphatic leukemia incidence and mortality was observed among a subgroup of employees who had exposure to polymer extrusion fumes, solvents, and colorants. However , no relationship was found with duration or estimated intensity of exposure to these agents. The authors were unable to establish an etiology for the leukemias, The current study updated the mortality among this cohort of workers another II years through 1986. During this updated period, the total number of deaths increased by more than twofold. Even so. mortality Table 7. Observed (0) and expected" (E) deaths from lyrnphatic and hematopoietic cancer , the standardized mortality ratio (SMR), and 95% confidence Intervals (95% CI) by mutually exclusive exposure and socioeconomic groupings. • Expected deaths based on United States mortality rates for white males. b Person-years were not accumulated for an employee until he satisfied the criteria of the grouping to which he was ultimately assigned. Thus the person-years and expected numbers of death's obtained by summing across all mutually exclusive groupings were slightly less than when summed independent of groupings. c SMR not calculated when fewer than three deaths were observed. 150 from all causes and total cancer remained below the expected levels, based on death rates of either the US male population or a comparable group of workers without exposure to styrene-based products. There were some statistically significant excesses in specific categories of death among particular subgroups of workers, and these deserve further discussion.
Mortality from symptoms, senility and ill-defined conditions was significantly elevated in the total cohort in comparison with that of the unexposed workers, and it was particularly increased among employees assigned to styrene monomer and finishing operations. Six of the seven deaths in this category occurred among employees from the Texas facility. The state of Texas is known to have an abnormally high death rate from symptoms, senility and ill-defined conditions, and this regional phenomenon appears to be the most likely explanation for the finding in this cohort (6).
A statistically significant excess of mortality from arteriosclerotic heart disease was observed among the employees who were categorized as having a potential for styrene and ethylbenzene exposures of~5 ppm. There was no evidence of a trend with duration of exposure, nor did the excess occur among workers while they were employed in jobs with potential exposure at these levels. Lacking a biological model or any support for this finding from other studies, it might reasonably be ascribed to chance. This is especiallytrue given the multiple comparisons that were done.
Given the a priori focus on mortality from lymphatic and hematopoietic cancer, a more-detailed discussion of the findings in this area is warranted. Mortality from each form of lymphatic and hemopoietic cancer was slightly elevated in the total cohort. Yet the excesses were not concentrated among the workers categorized as having the most intense exposures to styrene, nor were they concentrated among workers with longer term exposure in any of the areas. Instead the excesses were confined to men who had worked for relatively brief periods of time (ie, < 5 years) in polymerization, extrusion, and colorant operations. In particular, employees who had exposure to extrusion fumes and colorants showed the highest risks, especially from leukemia and multiple myeloma. This finding is in accord with the findings originally reported by Ott et al (1). Once again, the relatively small number of deaths and the multiple exposures of the workers make it impossible to isolate a particular agent as a likely cause of the excess, if indeed, any of these cases are related to the work environment.
The results of cohort studies conducted in related. but different, industries are important to consider. Table 8 summarizes seven published mortality studies (7)(8)(9)(10)(11)(12)(13) of workers who were potentially exposed to styrene-based products in the three general industry classifications of styrene-butadiene rubber, styrene monomer and polymerization, and reinforced plastics. It is critical to differentiate the exposure circumstances in each of these industrie s before consider ing their individual or collective results.
Two studies have been done of workers in the styrene-butadiene rubber industry (7,8). The ratio of 1,3-butadiene to styrene used to manufacture styrenebutadiene rubbe r was approximately three part s butadiene to one part styrene. No indust rial hygiene monitor ing data are available prior to the 1970s to estimate historical exposures to butadiene, styrene, or other chemicals (14). Surveys done since the middle 1970s have shown considerable variability in exposures across jobs within the industry. Typically, styrene and butadiene exposures have been measured below 5 ppm as an 8-h TWA, and benzene exposures have been below 1 ppm. Yet expo sures were likely to have been higher in the past, especially during World War II when the indust ry was getting sta rted and the demand for large production volumes was great.
Likewise, historical exposures in the styrene monomer and polymerization industry are largely unknown, but are likely to have been similar to tho se that existed Scand J Work Environ Health 1992, vol 18, no 3 in the styrene-butadiene rubber industry, except that exposures to butadiene monomer would have been minimal. Multiple concomitant exposures were common to both industries. The list of materials includes benzene and alkylbenzene compounds, styrene, formaldehyde, antio xidan ts, extender oils, acrylonitrile, organic dyes, and pigments.
By contrast , exposur es to styrene in the reinfor ced plastics industry have been much higher, ranging upwards to several hundred parts per million as recently as the 1970s (Ahlmark A: "Styrene Research Epidemiological Report"; prepared for the Swedish Plastics Federation, 1 October 1978). A much narrower spectrum of confounding agents was encountered, but did include glass fibers, acetone , methyl ethyl ketone, organic peroxides, and asbestos. For the aforementioned reasons, studies of workers in the reinforced plastics industry may be more suitable for addressing the potential hazards of styrene than studies in either the styrene-butadiene rubber or styrene monomer and polymerization industries. Almost without exception, each of the seven studies listed in table 8 reported mortality from all causes and total cancer among the cohorts to be at or below expected levels. Small numbers of observed and expected deaths from lymphatic and hematopoietic cancer plagues the interpretation of this outcome in six of the studies. An exception was the study by Matanoski et al (7), which combined data on 12160 workers from eight styrene-butadiene rubber plants. That study reported no increase in lymphatic and hematopoietic cancer mortality overall in the cohort. When the 9769 workers with complete work histories were assigned to one of four work areas based on the job they had held the longest, there was a statistically significant excess for black (6 observed, SMR 507,95% CI 187-1107), but not white (13 observed, SMR 110, 95070 CI 58-187), workers assigned to production. An examination of the subcategories showed statistically significant excess leukemia among black production workers (3 observed, SMR 656,95% CI 135-1906) and excess other lymphatic cancer among all production workers (9 observed, SMR 260, 95% CI 119-494). The nine deaths from other lymphatic cancer included four cases of non-Hodgkin's lymphoma and five of multiple myeloma.
No trend in mortality risk was apparent by duration of work or elapsed time since first employment in the styrene-butadiene rubber industry. The authors recommended that an assessment of estimated levels of exposure to butadiene and styrene be done for each job and used in a case-referent study (7). They then reported the preliminary results of such a study in which they found the risk of leukemia to be associated with exposure specificallyto butadiene (odds ratio 9.4, 95% CI 2.1-22.9), whereas there was no statistically significant increased risk associated with exposure to styrene (7). No other details of the case-referent study were reported, although Acquavella (15) has indicated that there was no validation of the death certificate diagnoses, and the exposure classification did not make use of available industrial hygiene monitoring data, nor did it consider differences in exposure potential across the eight plants. Additional research is being done on this cohort to address these criticisms (15).
Meinhardt et al (8) conducted a cohort study at two styrene-butadiene rubber plants not studied by Matanoski and her co-workers. Attention was originally directed at these two plants because of a 1976 report of two deaths from leukemia among the workers. Subsequent study of the 2756 white males with six or more months of employment at either of the two plants found a nonsignificant excess of lymphatic and hematopoietic cancer mortality among the workers from plant A (9 observed, SMR 155,95% CI 71-295), but not among the workers from plant B (2 observed, SMR 80, 95% CI 10-289). The majority of the excess mortality at plant A was attributable to an excess of leukemia (5 observed, SMR 203, 95% CI 66-474).

152
The authors noted that all five men with leukemia at plant A had worked in the period 1943-1945 for at least six months, prior to an operational change from batch to continuous polymerization. They then reanalyzed the data, restricting the cohort to those members meeting that condition. Not surprisingly, the SMR values for total lymphatic and hematopoietic cancer and leukemia were much increased in this subcohort. Lemen et al (16) recently reported that an update to the study by Meinhardt et al is underway.
Initial speculation had focused on styrene as a possible etiologic agent for the increased lymphatic and hematopoietic cancer mortality in the styrene-butadiene rubber industry. In recent years, toxicologic evidence has focused attention on the potential carcinogenic effect of 1,3-butadiene (17,18) rather than on that of styrene (19). A critical evaluation of the two studies conducted in the styrene-butadiene rubber industry (7,8) and additional studies of 1,3-butadiene monomer production workers (20,21) led one reviewer (14)to conclude that there was no convincing evidence linking adverse mortality effects to I ,3-butadiene exposure . This reviewer noted the poor quality of exposure data available and the absence of any relationship between duration of employment or latency and lymphatic and hematopoietic cancer mortality risk. By contrast, a second reviewer (22) examined the same three studies and concluded that they strongly suggested an etiologic association between occupational exposure to I ,3-butadiene and human cancer, specifically lymphatic and hematopoietic cancer. The second reviewer noted that the excess mortality in the three studies tended to be greatest among production and maintenance workers and was absent for office staff. According to this reviewer, the finding of excess mortality among workers exposed during World War II in two of the cohorts (8,20,21) was further evidence of a dose-response relationship. He also discounted the apparent inconsistency of subtypes of lymphatic and hematopoietic cancer reported in the different studies, citing evidence that indicates they may all derive from a common multipotential progenitor cell (23). Both reviewers (14,22) recommended that further efforts be made to quantify the butadiene exposures in the three cohorts, to permit firmer conclusions.
A subset of employees in the current study, particularly the group engaged in styrene-butadiene latex production, experienced 1,3-butadiene monomer exposures similar to those of workers in the styrenebutadiene rubber industry. A possible difference is that the latex product was maintained as an emulsion throughout the operation of the plants. Mortality from lymphatic and hematopoietic cancer was not excessive among the styrene-butadiene latex production workers in the current study (I observed, SMR 45,95% Cl 1-253). However, this subcohort of workers was small, and only a moderately increased risk could have been detected.
Interpretation of the two studies of workers in general styrene monomer production and polymerization operations (9,10) is similarly hampered by the small number of observed and expected deaths. Hodgson & Jones (9) reported statistically significant excess lymphatic and hematopoietic cancer mortality among workers producing styrene monomer and polymers (3 observed, SMR 536, 95070 CI 110-1565). All three of the deaths were from non-Hodgkin's lymphoma, although one was listed as Hodgkin's disease on the death certificate. Lymphatic leukemia was also reported for one man who was alive. Further analysis found no relationship between risk and length of service in styrene-exposed jobs. The authors noted that one man had worked in styrene-butadiene rubber production, another had had exposure to benzene, a third had had exposure to dyestuffs, antioxidants, and polyolefins, and the fourth had potential exposure to ethylene oxide.
Nicholson et al (10) reported one death each from malignant lymphoma and leukemia (2 observed, SMR 98, 95% CI 12-354) in a cohort of 550 unionized workers with at least five years of employment at a styrene monomer and polymerization facility . They concluded that the information was not definitive, except to indicate that any environmental risk of styrene exposure could not be extraordinary.
In the current study, the subcohort of employees engaged in the production of styrene monomer and polymers experienced exposures most similar to those of the workers in the two studies in the styrene monomer and polymerization industry. Small, nonstatistically significant excesses of mortality from lymphatic and hematopoietic cancer were observed among the employees assigned to styrene monomer and finishing production, or to polymerization, coloring, and extrusion operations. In the former group, the excess was confined to Hodgkin's disease (2 observed, SMR 526, 95% CI 59-1899), whereas in the latter the excess was particularly evident for multiple myeloma (5 observed, SMR 294, 95% CI 95-686) and leukemia (6 observed, SMR 165, 95% CI 60-359). In contrast to the study by Hodgson & Jones (9), non-Hodgkin's lymphoma mortality was not especially elevated in either group.
Other studies of styrene-exposed workers (11-13) have examined mortality in the reinforced plastics industry. None reported lymphatic and hematopoietic cancer or its specific variants to be elevated either overall or in subcohorts defined by duration of exposure, estimated intensity of exposure, or latency . In fact, each found a deficit of mortality, that reported by Coggon et al (10) reaching statistical significance (6 observed, SMR 40,95% CI 15-88). Thus, among workers from the industry with the highest documented exposures to styrene, there is no evidence of an increased risk of mortality from lymphatic and hematopoietic cancer. Th is observation is qualified by the fact that the reinforced plastics industry is relatively young, and consequently the length of follow-up of the workers is modest (ie, < 25 years). Continued monitoring of these three cohorts could provide more useful information .
With respect to the current study, the subcohort with styrene exposures which most closely parallel those in the reinforced plastics industry, the styrene-ethylben-zene~5 ppm TWA group, experienced four deaths from lymphatic and hematopoietic cancer compared with about three expected. It is difficult to draw conclusions based on such small numbers, but it would appear that there is no major risk associated with the potential for higher level styrene exposure in this cohort.
In summary, this update of mortality among employees who have manufactured styrene-based products continued to show the workers to have experienced lower than expected death rates from all causes and total cancer. This is in accord with other studies of workers with similar exposures in related, but different industries. An excess of mortality from lymphatic and hematopoietic cancer was observed for a subgroup of employees who held jobs which involved potential exposure to extrusion fumes, solvents, and colorants. Interpretation of this excess was complicated by the relatively small number of cases, their multiple chemical exposures, and possible confounding by life-style factors.
An excess of lymphatic and hematopoietic cancer mortality was not seen for the workers who produced styrene-butadiene latex, in contrast to recent reports of excess mortality among similarly exposed workers in the styrene-butadiene rubber industry. Nor was a statistically significant excess of mortality observed among the employees considered to have the highest potential exposures to styrene. This observation is consistent with findings from studies of workers in the reinforced plastics industry, which have failed to show excess lymphatic and hematopoietic cancer mortality associated with high-level exposure to styrene.