Document x1E8e1Nj0zpq5g61d1g9yYyNE

a./ Hfi- a3s'! kkkc Mortality Among Employees of PVC Fabricators Leonard Chiazze, Jr., Sc.D.; William E. Nichols, Ph.O.; and Otto Wong, Sc.D. A cross-sectional mortality study of 4,341 deaths occurring among current and former employees of 17 PVC fabricators during 1964-1973 is presented. The objectives are: (1) to iden tify any angiosarcoma deaths among the employees of these fabricators, and (2) to examine the distribution of deaths by cause. No angiosarcoma deaths were found among the study group. Sex-race-cause-specific Proportionate Mortality Ratios (PMR's) were computed, using the corresponding U.S. mor tality as the standard. Among white employees, there appears to be an excess in total cancer mortality, particularly that of the digestive system. Observed deaths were found to exceed the expected in cancers of the breast and urinary organs among white females. Deficit mortality was observed in cirrhosis of liver among both male and female white em ployees. The lanuary 1974 disclosure of three deaths from angio sarcoma of the liver in a single vinyl chloride polymerization plant and the attention focused on this new occupational disease are well known. A significant amount of information from animal ex perimentation and observation on humans engaged in the produc tion of vinyl chloride followed.1 However, there was no informa tion regarding the possible adverse effects on employees engaged in polyvinyl chloride (PVC) fabrication, where potential exposures were thought to be low and to result from the release of unreact ed monomer trapped in the resin. This issue was considered of great import because of the very large number of workers believed to be engaged in fabrication. In March 1974, representatives of PVC producers, who are members of the Organization Resources Counselors (ORC), Oc cupational Safety and Health Standards Croup, contacted ORC about the feasibility of a study of health risks to employees work ing with vinyl chloride and polyvinyl chloride. Since the National Institute for Occupational Safety and Health (NIOSH) was con ducting a study of vinyl chloride and resin producing employees, the scope of the proposed ORC study was limited to employees of From the Division of Biostatistics and Epidemiology, Department of Community Medicine and International Health. Georgetown University School of Medicine, Washington. DC 20007 (Drs. Chiazze and Wong) and Organization Resouces Coun selors. Inc., Washington. DC 20006 (Dr. Nichols). companies engaged in the fabrication of PVC resin into finished products. A number of alternative study designs was considered, and it was decided that a proportional mortality study would best meet the urgent need for information. The study concentrated on deaths occurring during the ten-year period 1964-1973 among ac tive fabricating employees plus retirees. Since it was not possible to identify those employees with only vinyl chloride exposure, the study focused on deceased employees who worked anywhere in those plants where PVC fabrication was carried on. The primary objective of the study was to determine whether or not any angio sarcoma deaths had occurred among employees of the fabricators under study. A secondary objective was to examine the distribu tion of all deaths by cause. The relatively few producers of vinyl chloride monomer and of PVC resin are generally medium to large companies. PVC fabri cators, however, range from large plants of major companies to very small job shops. Even in a large plant there may be few em ployees who are in the vicinity of PVC resins. The range of com pany plant size and number of exposed employees help to explain the wide range of estimates of total employment dependent upon PVC resins. Fabrication is geographically disbursed, with at least some operation in almost every state. Typical fabrication products include coated wire, upholstery fabrics, floor and wall coverings, pipe and other construction materials, toys, recreational equip ment, phonograph records, containers and container lining, and a myriad of novelty items. Since the fabrication of PVC resin into finished products is often a part of diversified product lines in a plant, raw materials other than resin are utilized. Thus, employees may be exposed to toxic agents other than vinyl chloride. Conse quently, in the current study, it is not possible to isolate only the effect of vinyl chloride. Some departures from expectation may be due to exposures to vinyl chloride, to one or more other agents, to some combination, or to employee characteristics other than occupation. Materials and Methods The decision to focus on a cross-sectional mortality study rather than on an historical cohort study derived from several con siderations. First, the primary study objective was to determine relatively quickly whether or not any angiosarcoma deaths could be identified among the study group, and this was best ac complished by examining causes of death among relatively recent Raprintad from Journal of Occupational Modicina Soptombor. 1977. Volumo 19. No. 9 pp. 923*029 BFG22046 22624001 1 (i.e., past ten years) decedents. The desirability of identifying an historical cohort for follow-up was recognized, but it was deter mined that it would not be possible to identify clearly and com pletely the necessary cohort of workers within a reasonable period of time, if at all. As a consequence, this study is based upon 4,341 deaths which occurred during the period 1964-1973 among current or former employees of 17 companies engaged in PVC fabrication. A total of 55 plants supplied data on all identifiable deaths since, as was mentioned earlier, it was not possible to identifv those employees with onlv vmvl chloride exposure for the studv. In order to be included in this study, a deceased employee fell into one of the following categories: (1) employee died while ac tively employed, (2) employee died after retiring from the com pany with retirement benefits, and (3) employee died after ter minating employment but with vesting in a company-sponsored life insurance plan. Deaths occurring among former employees with less than the number of vears necessary for insurance vesting were not included in plant records and were, therefore, not available for this study. An attempt was made to compensate for at least some of the "leakage" due to employees leaving the industry prior to the time they were vested in any insurance program by using the following procedures. Once death records for a given company were iden tified, death certificates were obtained and separated into two groups as follows: (1) all deaths with cancer (ICDA 140-205 up to 1967; ICDA 140-209 for 1968 on), or liver disease (ICDA 580-586 up to 1967; ICDA 570-576 for 1968 on), or suspected liver con dition mentioned on the death certificate; and (2) all deaths from causes other than cancer or liver disease. Death records men tioning cancer or liver disease and giving a hospital as place of death were segregated and sorted by hospital. Where there were several death records for a single hospital or several hospitals in the same area, arrangements were made for a Registered Records Administrator (RRA) to visit the hospitals, review medical records, contact the hospital pathologist and review pathology records to determine whether there were any angiosarcoma cases in the hospital's pathology records. If so, an attempt was made to deter mine whether or not the angiosarcoma possibly could be related to vinyl chloride exposure. This procedure enabled us to identify whether any employees, who had left the industry prior to vesting in an insurance program and remained in the local area, had died from angiosarcoma of the liver. The hospital inquiry procedures turned up five deaths from angiosarcoma of the liver; according to company employment records, none were employed at any time in PVC fabricating plants. These deaths were reported to OSHA, NIOSH, and CDC for inclusion in the nationwide angiosarcoma survey. Death certificates not mentioning a hospital as a place of death required an additional step to determine whether death occurred in a hospital or if a recent hospitalization had occurred prior to death. The certifying physician was queried by mail for in formation on recent hospitalization of the decedent and whether there was any indication that he or she had cancer or liver disease. Additionally identified hospitals were contacted as above. Originally, it was planned that death certificates would be ob tained from life insurance carriers. In practice, both company and life insurance records were used because of record retention prac tices and retrieval problems in the insurance carriers. All available death certificates were examined by a trained nosologist for men tion of angiosarcoma anywhere on the certificate, and no angio sarcomas were found on the death certificates. Underlying cause of death was classified according to the eighth revision of the In ternational Classification of Diseases.2 Over the ten-year period under study, plant size was found to have changed drastically in many cases. In one instance, a plant employing 300 employees in 1973 employed fewer than 10 in 1964. At the other extreme, one plant employing 5,000 people in 1964 had only 2,000 in 1973. Some companies had gone out of business. Such developments show not only the rapid changes taking place but also the difficulty of determining average industry employment. The 17 companies covered in this report ranged from single plant, small companies (one with 70 employees) to multiplant large companies (one with over 7,000 employees). Average number of employees clustered between 300 and 500. Although total employment is difficult to determine precisely, it is estimated to have been between 65,000 and 70,000 at the end of 1973. Since the population at risk could not be determined, mortality rates as measures of risk could not be calculated. Rather, results were summarized in terms of Proportionate Mortality Ratios (PMR) adjusted for the age distribution of the study group. The PMR uses relative frequencies of specific causes of death by age in a com parison population to obtain expected numbers of deaths from that cause in the study population. Selection of an appropriate comparison population poses some difficulties since it would be desirable to have a cause of death distribution for a similar group of workers dying during the same period, but not subjected to the factor under study. But such a comparison population was not available, and the study used mortality for the United States, specific for color and sex, for comparative purposes.3 Despite recognized deficiencies, the PMR can provide clues on unusual distributions of causes of death where true rates cannot be calculated. Results Table 1 gives the distribution of deaths among the employees of the 17 PVC fabricators within each of the ten years. An increase in the proportion of deaths taking place in later years of the decade may reflect a variety of factors, such as an increasing employee population, less complete availability of records for earlier years (e.g., in one company, there were no records for the first four of the ten years), increase in duration of exposure, etc. However, there is no reason to believe that any unavailable death records for earlier years would be concentrated in a few causes and thereby bias the distribution of deaths for those years. 22624002 BFG22047 occurred among employees of the PVC fabricators under study. Since no angiosarcoma deaths were found among the employees studied, the first question has an unequivocal answer. A secondary objective was that of examining the distribution of deaths by cause among the employees under study. Implicit in that objective is the question of whether or not that distribution is, in some sense, unusual. There is no unequivocal answer to the latter question. Whether or not an observed distribution of causes of death is unusual clearly relates to the standard or comparison population as well as the analytic methodology.4 On the basis of a proportionate mortality analysis, there appear to be excesses in total cancer mortality among both white men and white women in the study, when compared to the distributions of deaths for the total United States specific for color and sex and adjusted for age. Excesses in cancer mortality appear concentrated in cancers of the digestive system and, in particular, in cancers of the intestine for both men and women. In addition, there is a suggestion that mortality from cancer of the breast and urinary organs among white women employees is higher than that for the total U.S. There are, however, several reasons why definitive interpretation is difficult It has, for example, been suggested that a Proportionate Mortality Ratio based on an exter nal population may not be sufficiently discriminating in screening for potential hazards.5 In addition, an analysis using PMRs fails to take account of the absolute risk of dying in the population under study.6 Further, a number of studies have suggested an overall favorable mortality for industrial working populations, even in those where well-defined hazards increase risk for a specific cause.' * Factors such as these are meant to suggest that proportionate mortality analysis must be interpreted cautiously, with the intention of providing leads for further investigation. Results consistent with those of previously published studies would be of particular interest With no comparable working populations in the fabrication industry, potential contrasts may be found in studies of vinyl chloride workers. Monson, et al, provided a proportionate mortality analysis of 161 deceased workers (all presumably white males) in two plants, one where vinyl chloride monomer is produced and one where it is polymerized into polyvinyl chloride.6 Results of that study suggest a possible excess in total cancer mortality, primarily cancer of the BFG22048 \ M M TaW* Z. -- OMriMtaa rf OMftt Mum '-nr' * 17 PVC faMirtw by Cthr and In, 1964-1973. bn Total White Nonwhite Unknown Ttm4341 3*49 204 288 3676 3248 180 248 ta M 663 601 24 38 UMMW 2 0 0 2 The distribution of deaths in the study group by race and sex is given in Table 2. Most deaths (84.7%) occurred among males. The vast majority (88.7%) of decedents were white and there were 288 (6.6%) for whom race could not be determined. Table 3 shows this distribution of deaths by age. For both white male and white female employees, more than half of the deaths occurred among those aged 65 and over (58.1 % and 51.9% for white men and white women, respectively). The corresponding proportions of deaths which occurred after age 65 in the United States population in 1968 were 59% and 72%, respectively). The higher proportion of deaths among white women over 65 in the U.S. population compared to the employee population may indicate that relatively fewer women work to retirement age than do men. While the numbers of deaths among nonwhite employees are too small for any definite conclusions, the distribution of deaths by age among nonwhite men is roughly comparable to that for the total United States. Tables 4 and 5 show the distribution of deaths for selected causes by sex for whites and nonwhites, respectively. Among white men, nearly 60% of the deaths are from diseases of the cir culatory system (ICDA 390-458) and 20% are deaths from cancer (ICDA 140-209). The corresponding percentages for the total U.S. population of white males in 1968 (midpoint of the study period) were 54% for diseases of the circulatory system and 16% for can cer. Digestive and respiratory cancers account for somewhat less than two-thirds of all cancer deaths among the white males in the study group as compared to about 59% of all cancer deaths among U.S. white males in 1968. For white women, over 46% of the deaths are from diseases of the circulatory system (compared to 57% for all U.S. white women in 1968), and 30% were from can cer (compared to 18% for all U.S. white women in 1968). Somewhat over half of the cancer deaths among white women in the study were from cancers of the breast and digestive system, about the same proportion as for all U.S. white women in 1968. About two-thirds of the deaths for both nonwhite men and women were from diseases of the circulatory system and from cancer. The numbers of deaths for most causes among nonwhites, however, are quite small making interpretation difficult and are presented mainly for completeness. Age-adjusted PMRs for white male and white female employees of the 17 PVC fabricators are presented in Table 6 for selected causes of death. There were no similar analyses for nonwhites because for nearly all causes, the numbers of deaths were too small to produce stable PMRs. Expected numbers of deaths have been calculated on the basis of the sex-cause-age specific distribution of deaths among U.S. whites in 1968 applied to the total number of deaths by age among white men or white women in the study group. The midpoint of the study period (1968) was selected for the standard, since it is representative of the study period and provides U.S. mortality data coded according to the eighth revision of the International Classification of Diseases Ex pected numbers are deaths for individual causes to be expected in the study group if the proportion of total deaths ascribed to a given cause were the same as for the corresponding U.S. race-sex group, while accounting for differences between the age distributions of deaths in the study groups and deaths for the United States. A PMR larger than unity would indicate that the relative proportion of mortality from that particular cause in the study population is higher than would be expected based on the 1968 U.S. mortality experience. Due to the nature of the data, for mal significance tests and interpretation in probabilistic terms are deemed inappropriate. In addition to the numerical value of the PMR itself, the ob served number of deaths is also of interest The larger the number of deaths observed, the more reliable the PMR. There were two deaths of unknown age among the white males, one a digestive cancer (ICDA 153) and one an unspecified cancer (ICDA 199). These were included in the age groups with the lowest relative frequency for those causes, that is, less than 35 years for total can cer and digestive cancer and 65 and over for other and un specified cancer. Such an assignment has a trivial effect on the ex pectation but is the most conservative way of handling these unknowns, since it serves to increase the expectation (and hence lower the PMR) by the least amount. Based upon both the number of deaths and the value of the PMR, there seems to be an important excess in total cancer mor tality among the white males in the study group. The distribution by specific cancer site suggests that any excess appears to be con- 22624004 BFG22049 Tabt* 4. -- OtatrlbetiM af Dtatba fna SalKtad Cimm Amoitf Employ* of 17 PVC FaSrieaton by Sox, WMto (My, 1964-1973. Cmrss of OnM tCM Sth Rev. All Causes AH Cancers 140-209 Buccal Cavity and Pharyn* Digestive System Stomach Intestine Rectum liver Respiratory System Lung Bone. Connective Tissue. Skin. Breast Breast Genital Organs Urinary Organs Bram & Ottar Nervous System Other 6 Unspecified Lyptmmaa Leuhemiee Diaba-- Damn ot Hit Cirodatay Systaa Cerehroveaariv Otseest Ottawa of lew CboMtlMM. CMacyiMis and CMn|ili> Accidents, Poiienun. mi Viotmoe Acodants Suibda Ait Other Causes dtotdtafl 140 149 150159 151 153 154 155 160-163 162 170-174 174 180-187 188-189 191-192 190. 193-199 200-203. 206209 204-207 250 390458 430431 571 574-575 EUKM amm E950-C959 Me M Ne. 3249 666 % Nt. X 10000 20 50 601 181 10000 30 12 15 209 41 73 25 6 205 193 24 0 44 36 16 56 42 19 42 1931 299 42 046 643 126 225 077 018 6.31 594 3 53 8 24 8 0 12 12 050 882 133 3.99 133 0 200 200 0.74 -- 1.35 1.11 51 44 19 11 8.49 7 32 3.16 1.83 0.49 4 1 72 18 129 9 059 t 129 11 0.67 300 1.50 0.17 193 5945 281 4676 830 69 1099 129 3 099 7 022 3 090 m 5.79 43 799 134 415 932 40 123 2 033 m 11.46 7*. 1231 TaMd S. -- Matri--n af Dantba frtnt SahcM Clan Amo-- Ei--toy* of 17 PVC hbrfcatan by Sax. NmmM* (My. 1964-1673. Cm* of 0--k All Causes All Cancers Buccal Cavity and Pharyn* Digestive System Stomach Intestine Rectum liver Besoratorj Systam lung Bone. Connective Tissue. Skin. 8reast Brant Genital Organs Uritny Organs Bran 6 Other Nervous System Other 6 Unspecified lynyhames Lou--i Oiabttn Diseases ot the CimilHwy System Carabnmaaito Damn Ottawa cf Lmt ChgteMdmm dH--itb. 4 CM-- KM M Rdv. 140-209 140-149 150-159 151 153 154 155 160-163 162 170-174 174 180-187 189-189 191192 190. 193*196 200203. 201-209 204207 250 390458 430439 571 574679 Mi ne. % Ne. * 180 10000 24 100.00 39 21.66 8 33 33 0 0 00 13 722 4 16.67 3 167 0 0 2 1.11 1 4.16 1 0.56 0 0 1 056 0 0 7 389 0 0 7 389 0 0 0 0 3 1250 0 0 3 12.50 4 222 0 0 3 167 0 0 i 096 0 0 6 333 0 0 4 222 0 0 2 111 1 4.16 2 111 0 0 m 49.09 8 3393 19 1091 3 12.50 3 167 1 4.16 0 0 00 4 Wm Acbdetft M 01--Cans Emm E90M948 E99M9S9 33 1833 3 1294 17 9.44 i 4.16 2 in 0 0 22 12a 4 1667 centrated in digestive system cancer, given the large number of deaths (31% of all cancer deaths). For each of the sites within the digestive system, observed numbers of deaths are greater than ex pected. Although, except for intestinal cancer, the observed num bers of deaths are small. The PMR for liver cancer is rather high, out a definite conclusion is difficult with only six observed deaths. On the other hand, the PMR for cirrhosis of the liver suggests a deficit in mortality from that cause. For white women, the PMR for total cancer appears high, as it did for white men, with cancer of the digestive system and, perhaps, cancer of the breast contributing a fair amount to the ex cess. The PMRs for individual sites within the digestive system are based upon small numbers but suggest an increase in intestinal cancer. The PMR for urinary cancer, though based upon only 11 deaths, seems strikingly high and may be an indication for further investigation. On the other hand, in contrast to the observation in white men, the PMR for respiratory cancer among white women is very close to unity. Similar to the observation in white men, mor tality from cirrhosis of the liver seems to be in deficit, but the corresponding observed number of deaths is quite small. Among both white male and female employees, diseases of the circulatory system account for a large percentage of total deaths. In each case, observed numbers of deaths are close to expected. There appears to be a somewhat different pattern among men and women for deaths due to accidents, poisonings, violence with the number of deaths somewhat low among men and high among women. There were 39 cancer deaths among nonwhite men, somewhat greater than expectation, although the numbers by site are too small for meaningful analysis. Mortality from cerebrovascular disease appeared high (PMR - 2.11) but the finding was based upon only 19 deaths. Discussion The present study was designed with two objectives. The first was to determine whether or not any angiosarcoma deaths had BFG22050 22624005 liver. However, there was a suggestion that cancers of the lung and brain also appeared with excess frequency. The current study, while suggesting an excess in total cancer, particularly digestive cancer, does not clearly point to marked excesses for cancers of the liver, lung, and brain among white men, although the PMRs for these cancers are greater than one. The observed-to-expected PMRs reported by Monson for vascular lesions affecting the cen tral nervous system, circulatory diseases, and external causes are similar to those reported here. Tabershaw and Caffev. in a study of workers engaged in the manufacture of vinyl chloride and its polymers, conclude that no specific cause of death was statistically significantly greater than expectations based upon Standardized Mortality Ratios (SMRs) using the U S. male population as the standard.' By other criteria, however, the authors conclude there may be an excess risk for mortality from digestive cancer, respiratory cancer, cancer of other and unspecified sites, and lymphomas. At lower exposure levels, they suggest some excess for cancers of the buccal cavity and cancers of the other and unspecified sites The current study would seem to indicate that excesses of mor tality from cancer of the digestive system are not sex-specific and are not limited to liver cancer. The majority of the PMRs for cancer among both white men and white women are in excess of unity. Such results must be interpreted with caution but, since they ap pear to be consistent with previously studied workers, they suggest the need for continued investigation. Summary Results of a study of 4,341 deaths occurring among current and former employees of 17 PVC fabricators during the period 19641973 are presented. The study was carried out to determine whether any angiosarcoma deaths had occurred among em ployees of these fabricators and to examine the distribution of deaths by cause. No angiosarcoma deaths were found among the study population. Distributions by cause of death among white male and white female employees were compared with those for the entire U.S., specific for color and sex and adjusted for age by means of Proportionate Mortality Ratios. There appears to be an important excess in total cancer mortality among both white male and white female employees with digestive cancer, particularly that of the intestine, contributing to the excess. Among the women employees, cancers of the breast and urinary organs may also be in excess. No excesses were found for diseases of the cir culatory system. There were deficits in mortality from cirrhosis of the liver among both male and female employees and a deficit in accidents, poisonings, and violence among men. The use of the Proportionate Mortality Ratio based on an external standard requires caution in interpreting these results, but the need for fur ther investigations is indicated. This study was sponsored by the following organizations: American Can Company. 8. F. Goodrich Company. Diamond Shamrock Corporation. Firestone Tire and Rubber Company, lohns Manville, Motor Vehicle Manufacturers Association Stauffer Chemical Company, Union Carbide Corporation, Uniroval. Incorporated. References 1. SeUkott IJ and Hammond EC (Eds.): Toxicitv of vinvl chlonde -- polvvmvl chlonde Ann V Y Acad Sei 246:1975. 2. Manual of the International Classification of Diseases Adapted for Use in the United States. Eighth Revision. Washington, DC. U S. Covt. Printing Office, 1967. Vol. 1. 3. National Center For Health Statistics: Vital Statistics of the United States, 1968 - Volume II - Mortality, Part A. Washington, DC., U S. Covt. Printing Office. 1972. pp. 140-205. 4. Chiazze l.: Problems of study design and interpretation of industrial mortality experience. IOM 18:169-170. 1976. 5. Redmond CK and Breslin PP: Comparison of methods for assessing oc cupational hazards. IOM 17:313-317, 1975. 6. Monson RR, Peters |M. and lohnson MN: Proportional mortality among vinyl-chloride workers. Lancet 2:397-398. 1974. 7 Lloyd |W and Ciocco A: Long term mortality study of steelworkers: I Methodology IOM 11:299-310, 1969 8. Redmond CK. Ciocco A, Lloyd IW. and Rush HW: Long term mortality study of steelworkers: IV. Mortality from malignant neoplasms among coke oven workers. IOM 14:621-629. 1972. 9. Tabershaw IR and Gaffey WR: Mortality study of workers in the manufacture of vinyl chloride and its polymers. IOM 16:509-518, 1974. BFG22051 Oo 03